starred/windmill
1
0
Fork 0
mirror of https://github.com/windmill-labs/windmill.git synced 2026-08-19 16:02:14 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
586e178edd588975701dba714393fb19ae85ad57
windmill/backend/windmill-queue/src/jobs.rs
T

3870 lines
137 KiB
Rust
Raw Blame History

/*
* Author: Ruben Fiszel
* Copyright: Windmill Labs, Inc 2022
* This file and its contents are licensed under the AGPLv3 License.
* Please see the included NOTICE for copyright information and
* LICENSE-AGPL for a copy of the license.
*/
use std::{
borrow::Borrow,
collections::{HashMap, HashSet},
sync::Arc,
vec,
};
use anyhow::Context;
use async_recursion::async_recursion;
use axum::{
body::Bytes,
extract::{FromRequest, FromRequestParts, Query},
http::{request::Parts, Request, Uri},
response::{IntoResponse, Response},
};
use chrono::{DateTime, Duration, Utc};
use itertools::Itertools;
#[cfg(feature = "prometheus")]
use prometheus::IntCounter;
use regex::Regex;
use reqwest::{
header::{HeaderMap, CONTENT_TYPE},
Client, StatusCode,
};
use rsmq_async::RsmqConnection;
use serde::{Deserialize, Serialize};
use serde_json::{json, value::RawValue};
use sqlx::{types::Json, FromRow, Pool, Postgres, Transaction};
#[cfg(feature = "benchmark")]
use std::time::Instant;
use tokio::{sync::RwLock, time::sleep};
use tracing::{instrument, Instrument};
use ulid::Ulid;
use uuid::Uuid;
use windmill_audit::audit_ee::{audit_log, AuditAuthor};
use windmill_audit::ActionKind;
#[cfg(not(feature = "enterprise"))]
use windmill_common::worker::PriorityTags;
use windmill_common::{
db::{Authed, UserDB},
error::{self, to_anyhow, Error},
flow_status::{
BranchAllStatus, FlowCleanupModule, FlowStatus, FlowStatusModule, FlowStatusModuleWParent,
Iterator, JobResult, RestartedFrom, RetryStatus, MAX_RETRY_ATTEMPTS, MAX_RETRY_INTERVAL,
},
flows::{
add_virtual_items_if_necessary, FlowModule, FlowModuleValue, FlowValue, InputTransform,
},
jobs::{
get_payload_tag_from_prefixed_path, CompletedJob, JobKind, JobPayload, QueuedJob, RawCode,
},
oauth2::WORKSPACE_SLACK_BOT_TOKEN_PATH,
schedule::Schedule,
scripts::{get_full_hub_script_by_path, ScriptHash, ScriptLang},
users::{SUPERADMIN_NOTIFICATION_EMAIL, SUPERADMIN_SECRET_EMAIL},
utils::{not_found_if_none, report_critical_error, StripPath},
worker::{to_raw_value, DEFAULT_TAGS_PER_WORKSPACE, NO_LOGS, WORKER_CONFIG},
BASE_URL, DB, METRICS_ENABLED,
};
#[cfg(feature = "cloud")]
use windmill_common::users::SUPERADMIN_SYNC_EMAIL;
#[cfg(feature = "enterprise")]
use windmill_common::worker::CLOUD_HOSTED;
use crate::{
schedule::{get_schedule_opt, push_scheduled_job},
QueueTransaction,
};
#[cfg(feature = "prometheus")]
lazy_static::lazy_static! {
// TODO: these aren't synced, they should be moved into the queue abstraction once/if that happens.
static ref QUEUE_PUSH_COUNT: prometheus::IntCounter = prometheus::register_int_counter!(
"queue_push_count",
"Total number of jobs pushed to the queue."
)
.unwrap();
static ref QUEUE_DELETE_COUNT: prometheus::IntCounter = prometheus::register_int_counter!(
"queue_delete_count",
"Total number of jobs deleted from the queue."
)
.unwrap();
static ref QUEUE_PULL_COUNT: prometheus::IntCounter = prometheus::register_int_counter!(
"queue_pull_count",
"Total number of jobs pulled from the queue."
)
.unwrap();
pub static ref WORKER_EXECUTION_FAILED: Arc<RwLock<HashMap<String, IntCounter>>> = Arc::new(RwLock::new(HashMap::new()));
}
lazy_static::lazy_static! {
pub static ref HTTP_CLIENT: Client = reqwest::ClientBuilder::new()
.user_agent("windmill/beta")
.timeout(std::time::Duration::from_secs(20))
.connect_timeout(std::time::Duration::from_secs(10))
.build().unwrap();
}
#[cfg(feature = "cloud")]
const MAX_FREE_EXECS: i32 = 1000;
#[cfg(feature = "cloud")]
const MAX_FREE_CONCURRENT_RUNS: i32 = 30;
const ERROR_HANDLER_USERNAME: &str = "error_handler";
const SCHEDULE_ERROR_HANDLER_USERNAME: &str = "schedule_error_handler";
const SCHEDULE_RECOVERY_HANDLER_USERNAME: &str = "schedule_recovery_handler";
const ERROR_HANDLER_USER_GROUP: &str = "g/error_handler";
const ERROR_HANDLER_USER_EMAIL: &str = "error_handler@windmill.dev";
const SCHEDULE_ERROR_HANDLER_USER_EMAIL: &str = "schedule_error_handler@windmill.dev";
const SCHEDULE_RECOVERY_HANDLER_USER_EMAIL: &str = "schedule_recovery_handler@windmill.dev";
#[derive(Clone, Debug)]
pub struct CanceledBy {
pub username: Option<String>,
pub reason: Option<String>,
}
pub async fn cancel_single_job<'c>(
username: &str,
reason: Option<String>,
job_running: &QueuedJob,
w_id: &str,
mut tx: Transaction<'c, Postgres>,
db: &Pool<Postgres>,
rsmq: Option<rsmq_async::MultiplexedRsmq>,
force_cancel: bool,
) -> error::Result<(Transaction<'c, Postgres>, Option<Uuid>)> {
if ((job_running.running || job_running.root_job.is_some()) || (job_running.is_flow()))
&& !force_cancel
{
let id = sqlx::query_scalar!(
"UPDATE queue SET canceled = true, canceled_by = $1, canceled_reason = $2, scheduled_for = now(), suspend = 0 WHERE id = $3 AND workspace_id = $4 RETURNING id",
username,
reason,
job_running.id,
w_id
)
.fetch_optional(&mut *tx)
.await?;
if let Some(id) = id {
tracing::info!("Soft cancelling job {}", id);
}
} else {
let reason: String = reason
.clone()
.unwrap_or_else(|| "unexplicited reasons".to_string());
let e = serde_json::json!({"message": format!("Job canceled: {reason} by {username}"), "name": "Canceled", "reason": reason, "canceler": username});
append_logs(
job_running.id,
w_id.to_string(),
format!("canceled by {username}: (force cancel: {force_cancel})"),
db,
)
.await;
let add_job = add_completed_job_error(
&db,
job_running,
job_running.mem_peak.unwrap_or(0),
Some(CanceledBy { username: Some(username.to_string()), reason: Some(reason) }),
e,
rsmq.clone(),
"server",
false,
)
.await;
if let Err(e) = add_job {
tracing::error!("Failed to add canceled job: {}", e);
}
}
if let Some(mut rsmq) = rsmq.clone() {
rsmq.change_message_visibility(&job_running.tag, &job_running.id.to_string(), 0)
.await
.map_err(|e| anyhow::anyhow!(e))?;
}
Ok((tx, Some(job_running.id)))
}
pub async fn cancel_job<'c>(
username: &str,
reason: Option<String>,
id: Uuid,
w_id: &str,
mut tx: Transaction<'c, Postgres>,
db: &Pool<Postgres>,
rsmq: Option<rsmq_async::MultiplexedRsmq>,
force_cancel: bool,
) -> error::Result<(Transaction<'c, Postgres>, Option<Uuid>)> {
let job = get_queued_job_tx(id, &w_id, &mut tx).await?;
if job.is_none() {
return Ok((tx, None));
}
let job = job.unwrap();
// get all children
let mut jobs = vec![id];
let mut jobs_to_cancel = vec![];
while !jobs.is_empty() {
let p_job = jobs.pop();
let new_jobs = sqlx::query_scalar!(
"SELECT id FROM queue WHERE parent_job = $1 AND workspace_id = $2",
p_job,
w_id
)
.fetch_all(&mut *tx)
.await?;
jobs.extend(new_jobs.clone());
jobs_to_cancel.extend(new_jobs);
}
let (ntx, _) = cancel_single_job(
username,
reason.clone(),
&job,
w_id,
tx,
db,
rsmq.clone(),
force_cancel,
)
.await?;
tx = ntx;
// soft cancel parent if force cancel and job is a flow step
if force_cancel && job.is_flow_step {
if let Some(parent_job_id) = job.parent_job {
let job = get_queued_job_tx(parent_job_id, &w_id, &mut tx).await?;
if let Some(job) = job {
let (ntx, _) = cancel_single_job(
username,
reason.clone(),
&job,
w_id,
tx,
db,
rsmq.clone(),
false,
)
.await?;
tx = ntx;
}
}
}
// cancel children
for job_id in jobs_to_cancel {
let job = get_queued_job_tx(job_id, &w_id, &mut tx).await?;
if let Some(job) = job {
let (ntx, _) = cancel_single_job(
username,
reason.clone(),
&job,
w_id,
tx,
db,
rsmq.clone(),
force_cancel,
)
.await?;
tx = ntx;
}
}
Ok((tx, Some(id)))
}
/* TODO retry this? */
#[tracing::instrument(level = "trace", skip_all)]
pub async fn append_logs(
job_id: uuid::Uuid,
workspace: impl AsRef<str>,
logs: impl AsRef<str>,
db: impl Borrow<Pool<Postgres>>,
) {
if logs.as_ref().is_empty() {
return;
}
if *NO_LOGS {
tracing::info!("NO LOGS [{job_id}]: {}", logs.as_ref());
return;
}
if let Err(err) = sqlx::query!(
"INSERT INTO job_logs (logs, job_id, workspace_id) VALUES ($1, $2, $3) ON CONFLICT (job_id) DO UPDATE SET logs = concat(job_logs.logs, $1::text)",
logs.as_ref(),
job_id,
workspace.as_ref(),
)
.execute(db.borrow())
.await
{
tracing::error!(%job_id, %err, "error updating logs for large_log job {job_id}: {err}");
}
}
pub async fn cancel_persistent_script_jobs<'c>(
username: &str,
reason: Option<String>,
script_path: &str,
w_id: &str,
db: &Pool<Postgres>,
rsmq: Option<rsmq_async::MultiplexedRsmq>,
) -> error::Result<Vec<Uuid>> {
// There can be only one perpetual script run per 10 seconds window. We execute the cancel twice with 5s interval
// to avoid the case of a job _about to be restarted_ when the first cancel is run.
let cancelled_job_ids_first_batch = cancel_persistent_script_jobs_internal(
username,
reason.clone(),
script_path,
w_id,
db,
rsmq.clone(),
)
.await?;
sleep(std::time::Duration::from_secs(5)).await;
let cancelled_job_ids_second_batch = cancel_persistent_script_jobs_internal(
username,
reason.clone(),
script_path,
w_id,
db,
rsmq.clone(),
)
.await?;
return Ok(cancelled_job_ids_first_batch
.into_iter()
.chain(cancelled_job_ids_second_batch)
.collect_vec());
}
async fn cancel_persistent_script_jobs_internal<'c>(
username: &str,
reason: Option<String>,
script_path: &str,
w_id: &str,
db: &Pool<Postgres>,
rsmq: Option<rsmq_async::MultiplexedRsmq>,
) -> error::Result<Vec<Uuid>> {
let mut tx = db.begin().await?;
// we could have retrieved the job IDs in the first query where we retrieve the hashes, but just in case a job was inserted in the queue right in-between the two above query, we re-do the fetch here
let jobs_to_cancel = sqlx::query_scalar::<_, Uuid>(
"SELECT id FROM queue WHERE workspace_id = $1 AND script_path = $2 AND canceled = false",
)
.bind(w_id)
.bind(script_path)
.fetch_all(&mut *tx)
.await?;
// Then we cancel all the jobs currently in the queue, one by one
for queued_job_id in jobs_to_cancel.clone() {
let (new_tx, _) = cancel_job(
username,
reason.clone(),
queued_job_id,
w_id,
tx,
db,
rsmq.clone(),
false,
)
.await?;
tx = new_tx;
}
tx.commit().await?;
return Ok(jobs_to_cancel);
}
#[derive(Serialize, Debug)]
pub struct WrappedError {
pub error: serde_json::Value,
}
pub trait ValidableJson {
fn is_valid_json(&self) -> bool;
}
impl ValidableJson for WrappedError {
fn is_valid_json(&self) -> bool {
true
}
}
impl ValidableJson for Box<RawValue> {
fn is_valid_json(&self) -> bool {
!self.get().is_empty()
}
}
impl ValidableJson for serde_json::Value {
fn is_valid_json(&self) -> bool {
true
}
}
impl<T: ValidableJson> ValidableJson for Json<T> {
fn is_valid_json(&self) -> bool {
self.0.is_valid_json()
}
}
pub async fn register_metric<T, F, F2, R>(
l: &Arc<RwLock<HashMap<String, T>>>,
s: &str,
fnew: F2,
f: F,
) -> Option<R>
where
F: FnOnce(&T) -> R,
F2: FnOnce(&str) -> (T, R),
{
if METRICS_ENABLED.load(std::sync::atomic::Ordering::Relaxed) {
let lock = l.read().await;
let counter = lock.get(s);
if let Some(counter) = counter {
let r = f(counter);
drop(lock);
Some(r)
} else {
drop(lock);
let (metric, r) = fnew(s);
let mut m = l.write().await;
(*m).insert(s.to_string(), metric);
Some(r)
}
} else {
None
}
}
#[instrument(level = "trace", skip_all)]
pub async fn add_completed_job_error<R: rsmq_async::RsmqConnection + Clone + Send>(
db: &Pool<Postgres>,
queued_job: &QueuedJob,
mem_peak: i32,
canceled_by: Option<CanceledBy>,
e: serde_json::Value,
rsmq: Option<R>,
_worker_name: &str,
flow_is_done: bool,
) -> Result<WrappedError, Error> {
#[cfg(feature = "prometheus")]
register_metric(
&WORKER_EXECUTION_FAILED,
&queued_job.tag,
|s| {
let counter = prometheus::register_int_counter!(prometheus::Opts::new(
"worker_execution_failed",
"Number of jobs having failed"
)
.const_label("name", _worker_name)
.const_label("tag", s))
.expect("register prometheus metric");
counter.inc();
(counter, ())
},
|c| c.inc(),
)
.await;
let result = WrappedError { error: e };
tracing::error!(
"job {} did not succeed: {}",
queued_job.id,
serde_json::to_string(&result).unwrap_or_else(|_| "".to_string())
);
let _ = add_completed_job(
db,
&queued_job,
false,
false,
Json(&result),
mem_peak,
canceled_by,
rsmq,
flow_is_done,
)
.await?;
Ok(result)
}
lazy_static::lazy_static! {
pub static ref GLOBAL_ERROR_HANDLER_PATH_IN_ADMINS_WORKSPACE: Option<String> = std::env::var("GLOBAL_ERROR_HANDLER_PATH_IN_ADMINS_WORKSPACE").ok();
}
#[instrument(level = "trace", skip_all, name = "add_completed_job")]
pub async fn add_completed_job<
T: Serialize + Send + Sync + ValidableJson,
R: rsmq_async::RsmqConnection + Clone + Send,
>(
db: &Pool<Postgres>,
queued_job: &QueuedJob,
success: bool,
skipped: bool,
result: Json<&T>,
mem_peak: i32,
canceled_by: Option<CanceledBy>,
rsmq: Option<R>,
flow_is_done: bool,
) -> Result<Uuid, Error> {
// tracing::error!("Start");
// let start = tokio::time::Instant::now();
if !result.is_valid_json() {
return Err(Error::InternalErr(
"Result of job is invalid json (empty)".to_string(),
));
}
let mut tx: QueueTransaction<'_, R> = (rsmq.clone(), db.begin().await?).into();
let job_id = queued_job.id;
// tracing::error!("1 {:?}", start.elapsed());
tracing::debug!(
"completed job {} {}",
queued_job.id,
serde_json::to_string(&result).unwrap_or_else(|_| "".to_string())
);
let mem_peak = mem_peak.max(queued_job.mem_peak.unwrap_or(0));
let _duration: i64 = sqlx::query_scalar!(
"INSERT INTO completed_job AS cj
( workspace_id
, id
, parent_job
, created_by
, created_at
, started_at
, duration_ms
, success
, script_hash
, script_path
, args
, result
, raw_code
, raw_lock
, canceled
, canceled_by
, canceled_reason
, job_kind
, schedule_path
, permissioned_as
, flow_status
, raw_flow
, is_flow_step
, is_skipped
, language
, email
, visible_to_owner
, mem_peak
, tag
, priority
)
VALUES ($1, $2, $3, $4, $5, COALESCE($6, now()), (EXTRACT('epoch' FROM (now())) - EXTRACT('epoch' FROM (COALESCE($6, now()))))*1000, $7, $8, $9,\
$10, $11, $12, $13, $14, $15, $16, $17, $18, $19, $20, $21, $22, $23, $24, $25, $26, $27, $28, $29)
ON CONFLICT (id) DO UPDATE SET success = $7, result = $11 RETURNING duration_ms",
queued_job.workspace_id,
queued_job.id,
queued_job.parent_job,
queued_job.created_by,
queued_job.created_at,
queued_job.started_at,
success,
queued_job.script_hash.map(|x| x.0),
queued_job.script_path,
&queued_job.args as &Option<Json<HashMap<String, Box<RawValue>>>>,
result as Json<&T>,
queued_job.raw_code,
queued_job.raw_lock,
canceled_by.is_some(),
canceled_by.clone().map(|cb| cb.username).flatten(),
canceled_by.clone().map(|cb| cb.reason).flatten(),
queued_job.job_kind.clone() as JobKind,
queued_job.schedule_path,
queued_job.permissioned_as,
&queued_job.flow_status as &Option<Json<Box<RawValue>>>,
&queued_job.raw_flow as &Option<Json<Box<RawValue>>>,
queued_job.is_flow_step,
skipped,
queued_job.language.clone() as Option<ScriptLang>,
queued_job.email,
queued_job.visible_to_owner,
if mem_peak > 0 { Some(mem_peak) } else { None },
queued_job.tag,
queued_job.priority,
)
.fetch_one(&mut tx)
.await
.map_err(|e| Error::InternalErr(format!("Could not add completed job {job_id}: {e:#}")))?;
// tracing::error!("2 {:?}", start.elapsed());
if !queued_job.is_flow_step {
if _duration > 500
&& (queued_job.job_kind == JobKind::Script || queued_job.job_kind == JobKind::Preview)
{
if let Err(e) = sqlx::query!(
"UPDATE completed_job SET flow_status = q.flow_status FROM queue q WHERE completed_job.id = $1 AND q.id = $1 AND q.workspace_id = $2 AND completed_job.workspace_id = $2 AND q.flow_status IS NOT NULL",
&queued_job.id,
&queued_job.workspace_id
)
.execute(&mut tx)
.await {
tracing::error!("Could not update job duration: {}", e);
}
}
if let Some(parent_job) = queued_job.parent_job {
if let Err(e) = sqlx::query_scalar!(
"UPDATE queue SET flow_status = jsonb_set(jsonb_set(COALESCE(flow_status, '{}'::jsonb), array[$1], COALESCE(flow_status->$1, '{}'::jsonb)), array[$1, 'duration_ms'], to_jsonb($2::bigint)) WHERE id = $3 AND workspace_id = $4",
&queued_job.id.to_string(),
_duration,
parent_job,
&queued_job.workspace_id
)
.execute(&mut tx)
.await {
tracing::error!("Could not update parent job flow_status: {}", e);
}
}
}
// tracing::error!("Added completed job {:#?}", queued_job);
let mut skip_downstream_error_handlers = false;
tx = delete_job(tx, &queued_job.workspace_id, job_id).await?;
// tracing::error!("3 {:?}", start.elapsed());
if queued_job.is_flow_step {
if let Some(parent_job) = queued_job.parent_job {
// persist the flow last progress timestamp to avoid zombie flow jobs
tracing::debug!(
"Persisting flow last progress timestamp to flow job: {:?}",
parent_job
);
sqlx::query!(
"UPDATE queue SET last_ping = now() WHERE id = $1 AND workspace_id = $2",
parent_job,
&queued_job.workspace_id
)
.execute(&mut tx)
.await?;
if flow_is_done {
let r = sqlx::query_scalar!(
"UPDATE parallel_monitor_lock SET last_ping = now() WHERE parent_flow_id = $1 and job_id = $2 RETURNING 1",
parent_job,
&queued_job.id
).fetch_optional(&mut tx).await?;
if r.is_some() {
tracing::info!(
"parallel flow is done, setting parallel monitor last ping lock for job {}",
&queued_job.id
);
}
}
}
} else {
if queued_job.schedule_path.is_some() && queued_job.script_path.is_some() {
let schedule_path = queued_job.schedule_path.as_ref().unwrap();
let script_path = queued_job.script_path.as_ref().unwrap();
let schedule = get_schedule_opt(
tx.transaction_mut(),
&queued_job.workspace_id,
schedule_path,
)
.await?;
if let Some(schedule) = schedule {
skip_downstream_error_handlers = schedule.ws_error_handler_muted;
// script or flow that failed on start and might not have been rescheduled
let schedule_next_tick = !queued_job.is_flow()
|| {
let flow_status = queued_job.parse_flow_status();
flow_status.is_some_and(|fs| {
fs.step == 0
&& fs.modules.get(0).is_some_and(|m| {
matches!(m, FlowStatusModule::WaitingForPriorSteps { .. }) || matches!(m, FlowStatusModule::Failure { job, ..} if job == &Uuid::nil())
})
})
};
if schedule_next_tick {
if let Err(err) = handle_maybe_scheduled_job(
rsmq.clone(),
db,
queued_job,
&schedule,
script_path,
&queued_job.workspace_id,
)
.await
{
match err {
Error::QuotaExceeded(_) => return Err(err.into()),
// scheduling next job failed and could not disable schedule => make zombie job to retry
_ => return Ok(job_id),
}
};
}
if let Err(err) = apply_schedule_handlers(
rsmq.clone(),
db,
&schedule,
script_path,
&queued_job.workspace_id,
success,
result,
job_id,
queued_job.started_at.unwrap_or(chrono::Utc::now()),
queued_job.priority,
)
.await
{
if !success {
tracing::error!("Could not apply schedule error handler: {}", err);
let base_url = BASE_URL.read().await;
let w_id: &String = &queued_job.workspace_id;
if !matches!(err, Error::QuotaExceeded(_)) {
report_error_to_workspace_handler_or_critical_side_channel(
rsmq.clone(),
&queued_job,
db,
format!(
"Failed to push schedule error handler job to handle failed job ({base_url}/run/{}?workspace={w_id}): {}",
queued_job.id,
err
),
)
.await;
}
} else {
tracing::error!("Could not apply schedule recovery handler: {}", err);
}
};
} else {
tracing::error!(
"Schedule {schedule_path} in {} not found. Impossible to schedule again and apply schedule handlers",
&queued_job.workspace_id
);
}
}
}
if queued_job.concurrent_limit.is_some() {
let concurrency_key = match concurrency_key(db, queued_job).await {
Ok(c) => c,
Err(e) => {
tracing::error!(
"Could not get concurrency key for job {} defaulting to default key: {e:?}",
queued_job.id
);
legacy_concurrency_key(db, queued_job)
.await
.unwrap_or_else(|| queued_job.full_path_with_workspace())
}
};
if let Err(e) = sqlx::query_scalar!(
"UPDATE concurrency_counter SET job_uuids = job_uuids - $2 WHERE concurrency_id = $1",
concurrency_key,
queued_job.id.hyphenated().to_string(),
)
.execute(&mut tx)
.await
{
tracing::error!("Could not decrement concurrency counter: {}", e);
}
if let Err(e) = sqlx::query_scalar!(
"UPDATE concurrency_key SET ended_at = now() WHERE job_id = $1",
queued_job.id,
)
.execute(&mut tx)
.await
.map_err(|e| {
Error::InternalErr(format!(
"Error updating to add ended_at timestamp concurrency_key={concurrency_key}: {e:#}"
))
}) {
tracing::error!("Could not update concurrency_key: {}", e);
}
tracing::debug!("decremented concurrency counter");
}
tx.commit().await?;
tracing::info!(
%job_id,
root_job = ?queued_job.root_job.map(|x| x.to_string()).unwrap_or_else(|| String::new()),
path = &queued_job.script_path(),
job_kind = ?queued_job.job_kind,
started_at = ?queued_job.started_at.map(|x| x.to_string()).unwrap_or_else(|| String::new()),
duration = ?_duration,
permissioned_as = ?queued_job.permissioned_as,
email = ?queued_job.email,
created_by = queued_job.created_by,
is_flow_step = queued_job.is_flow_step,
language = ?queued_job.language,
success,
"inserted completed job: {} (success: {success})",
queued_job.id
);
#[cfg(feature = "cloud")]
if *CLOUD_HOSTED && !queued_job.is_flow() && _duration > 1000 {
let additional_usage = _duration / 1000;
let w_id = &queued_job.workspace_id;
let premium_workspace =
sqlx::query_scalar!("SELECT premium FROM workspace WHERE id = $1", w_id)
.fetch_one(db)
.await
.map_err(|e| Error::InternalErr(format!("fetching if {w_id} is premium: {e:#}")))?;
let _ = sqlx::query!(
"INSERT INTO usage (id, is_workspace, month_, usage)
VALUES ($1, TRUE, EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date), $2)
ON CONFLICT (id, is_workspace, month_) DO UPDATE SET usage = usage.usage + $2",
w_id,
additional_usage as i32)
.execute(db)
.await
.map_err(|e| Error::InternalErr(format!("updating usage: {e:#}")));
if !premium_workspace {
let _ = sqlx::query!(
"INSERT INTO usage (id, is_workspace, month_, usage)
VALUES ($1, FALSE, EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date), $2)
ON CONFLICT (id, is_workspace, month_) DO UPDATE SET usage = usage.usage + $2",
queued_job.email,
additional_usage as i32)
.execute(db)
.await
.map_err(|e| Error::InternalErr(format!("updating usage: {e:#}")));
}
}
#[cfg(feature = "enterprise")]
if !success {
if queued_job.email == ERROR_HANDLER_USER_EMAIL {
let base_url = BASE_URL.read().await;
let w_id = &queued_job.workspace_id;
report_critical_error(format!(
"Workspace error handler job failed ({base_url}/run/{}?workspace={w_id}){}",
queued_job.id,
queued_job
.parent_job
.map(|id| format!(
" trying to handle failed job ({base_url}/run/{id}?workspace={w_id})"
))
.unwrap_or("".to_string()),
))
.await;
} else if queued_job.email == SCHEDULE_ERROR_HANDLER_USER_EMAIL {
let base_url = BASE_URL.read().await;
let w_id = &queued_job.workspace_id;
report_error_to_workspace_handler_or_critical_side_channel(
rsmq.clone(),
&queued_job,
db,
format!(
"Schedule error handler job failed ({base_url}/run/{}?workspace={w_id}){}",
queued_job.id,
queued_job
.parent_job
.map(|id| format!(
" trying to handle failed job: {base_url}/run/{id}?workspace={w_id}"
))
.unwrap_or("".to_string()),
),
)
.await;
} else if !skip_downstream_error_handlers
&& matches!(queued_job.job_kind, JobKind::Flow | JobKind::Script)
&& queued_job.parent_job.is_none()
{
let result = serde_json::from_str(
&serde_json::to_string(result.0).unwrap_or_else(|_| "{}".to_string()),
)
.unwrap_or_else(|_| json!({}));
let result = if result.is_object() || result.is_null() {
result
} else {
json!({ "error": result })
};
tracing::info!(
"Sending error of job {} to error handlers (if any)",
queued_job.id
);
if let Err(e) =
send_error_to_global_handler(rsmq.clone(), &queued_job, db, Json(&result)).await
{
tracing::error!(
"Could not run global error handler for job {}: {}",
&queued_job.id,
e
);
}
if let Err(err) = send_error_to_workspace_handler(
rsmq.clone(),
&queued_job,
canceled_by.is_some(),
db,
Json(&result),
)
.await
{
match err {
Error::QuotaExceeded(_) => {}
_ => {
let base_url = BASE_URL.read().await;
let w_id: &String = &queued_job.workspace_id;
report_critical_error(format!(
"Could not push workspace error handler for failed job ({base_url}/run/{}?workspace={w_id}): {}",
queued_job.id,
err
))
.await;
}
}
}
}
}
if !queued_job.is_flow_step && queued_job.job_kind == JobKind::Script && canceled_by.is_none() {
if let Some(hash) = queued_job.script_hash {
let p = sqlx::query_scalar!(
"SELECT restart_unless_cancelled FROM script WHERE hash = $1 AND workspace_id = $2",
hash.0,
&queued_job.workspace_id
)
.fetch_optional(db)
.await?
.flatten()
.unwrap_or(false);
if p {
let tx = PushIsolationLevel::IsolatedRoot(db.clone(), rsmq);
// perpetual jobs can run one job per 10s max. If the job was faster than 10s, schedule the next one with the appropriate delay
let now = chrono::Utc::now();
let scheduled_for = if now
.signed_duration_since(queued_job.started_at.unwrap_or(now))
.num_seconds()
< 10
{
let next_run = queued_job.started_at.unwrap_or(now)
+ chrono::Duration::try_seconds(10).unwrap();
tracing::warn!("Perpetual script {:?} is running too fast, only 1 job per 10s it supported. Scheduling next run for {:?}", queued_job.script_path, next_run);
Some(next_run)
} else {
None
};
let (_uuid, tx) = push(
db,
tx,
&queued_job.workspace_id,
JobPayload::ScriptHash {
hash,
path: queued_job.script_path().to_string(),
custom_concurrency_key: custom_concurrency_key(db, queued_job.id).await?,
concurrent_limit: queued_job.concurrent_limit,
concurrency_time_window_s: queued_job.concurrency_time_window_s,
cache_ttl: queued_job.cache_ttl,
dedicated_worker: None,
language: queued_job
.language
.clone()
.unwrap_or_else(|| ScriptLang::Deno),
priority: queued_job.priority,
},
queued_job
.args
.as_ref()
.map(|x| PushArgs { args: x.0.clone(), extra: HashMap::new() })
.unwrap_or_else(PushArgs::empty),
&queued_job.created_by,
&queued_job.email,
queued_job.permissioned_as.clone(),
scheduled_for,
queued_job.schedule_path.clone(),
None,
None,
None,
false,
false,
None,
queued_job.visible_to_owner,
Some(queued_job.tag.clone()),
queued_job.timeout,
None,
queued_job.priority,
)
.await?;
if let Err(e) = tx.commit().await {
tracing::error!("Could not restart job {}: {}", queued_job.id, e);
}
}
}
}
// tracing::error!("4 {:?}", start.elapsed());
Ok(queued_job.id)
}
pub async fn send_error_to_global_handler<
'a,
T: Serialize + Send + Sync,
R: rsmq_async::RsmqConnection + Clone + Send,
>(
rsmq: Option<R>,
queued_job: &QueuedJob,
db: &Pool<Postgres>,
result: Json<&T>,
) -> Result<(), Error> {
if let Some(ref global_error_handler) = *GLOBAL_ERROR_HANDLER_PATH_IN_ADMINS_WORKSPACE {
let prefixed_global_error_handler_path = if global_error_handler.starts_with("script/")
|| global_error_handler.starts_with("flow/")
{
global_error_handler.clone()
} else {
format!("script/{}", global_error_handler)
};
push_error_handler(
db,
rsmq,
queued_job.id,
queued_job.schedule_path.clone(),
queued_job.script_path.clone(),
queued_job.is_flow(),
&queued_job.workspace_id,
&prefixed_global_error_handler_path,
result,
None,
queued_job.started_at,
None,
&queued_job.email,
false,
true,
None,
)
.await?;
}
Ok(())
}
pub async fn report_error_to_workspace_handler_or_critical_side_channel<
R: rsmq_async::RsmqConnection + Clone + Send,
>(
rsmq: Option<R>,
queued_job: &QueuedJob,
db: &Pool<Postgres>,
error_message: String,
) -> () {
let w_id = &queued_job.workspace_id;
let (error_handler, error_handler_extra_args) = sqlx::query_as::<_, (Option<String>, Option<Json<Box<RawValue>>>)>(
"SELECT error_handler, error_handler_extra_args FROM workspace_settings WHERE workspace_id = $1",
).bind(&w_id)
.fetch_optional(db)
.await
.ok()
.flatten()
.unwrap_or((None, None));
if let Some(error_handler) = error_handler {
if let Err(err) = push_error_handler(
db,
rsmq,
queued_job.id,
queued_job.schedule_path.clone(),
queued_job.script_path.clone(),
queued_job.is_flow(),
w_id,
&error_handler,
Json(&json!({
"error": {
"message": error_message
}
})),
None,
queued_job.started_at,
error_handler_extra_args,
&queued_job.email,
false,
false,
None,
)
.await
{
tracing::error!(
"Could not push workspace error handler trying to handle critical error ({error_message}) of failed job ({}): {}",
queued_job.id,
err
);
report_critical_error(error_message).await;
}
} else {
report_critical_error(error_message).await;
}
}
pub async fn send_error_to_workspace_handler<
'a,
'c,
T: Serialize + Send + Sync,
R: rsmq_async::RsmqConnection + Clone + Send,
>(
rsmq: Option<R>,
queued_job: &QueuedJob,
is_canceled: bool,
db: &Pool<Postgres>,
result: Json<&'a T>,
) -> Result<(), Error> {
let w_id = &queued_job.workspace_id;
let (error_handler, error_handler_extra_args, error_handler_muted_on_cancel) = sqlx::query_as::<_, (Option<String>, Option<Json<Box<RawValue>>>, bool)>(
"SELECT error_handler, error_handler_extra_args, error_handler_muted_on_cancel FROM workspace_settings WHERE workspace_id = $1",
).bind(&w_id)
.fetch_optional(db)
.await
.context("fetching error handler info from workspace_settings")?
.ok_or_else(|| Error::InternalErr(format!("no workspace settings for id {w_id}")))?;
if is_canceled && error_handler_muted_on_cancel {
return Ok(());
}
if let Some(error_handler) = error_handler {
let ws_error_handler_muted: Option<bool> = match queued_job.job_kind {
JobKind::Script => {
sqlx::query_scalar!(
"SELECT ws_error_handler_muted FROM script WHERE workspace_id = $1 AND hash = $2",
queued_job.workspace_id,
queued_job.script_hash.unwrap().0,
)
.fetch_optional(db)
.await?
}
JobKind::Flow => {
sqlx::query_scalar!(
"SELECT ws_error_handler_muted FROM flow WHERE workspace_id = $1 AND path = $2",
queued_job.workspace_id,
queued_job.script_path.as_ref().unwrap(),
)
.fetch_optional(db)
.await?
}
_ => None,
};
let muted = ws_error_handler_muted.unwrap_or(false);
if !muted {
tracing::info!("workspace error handled for job {}", &queued_job.id);
push_error_handler(
db,
rsmq,
queued_job.id,
queued_job.schedule_path.clone(),
queued_job.script_path.clone(),
queued_job.is_flow(),
&queued_job.workspace_id,
&error_handler,
result,
None,
queued_job.started_at,
error_handler_extra_args,
&queued_job.email,
false,
false,
None,
)
.await?;
}
}
Ok(())
}
#[instrument(level = "trace", skip_all)]
pub async fn handle_maybe_scheduled_job<'c, R: rsmq_async::RsmqConnection + Clone + Send + 'c>(
rsmq: Option<R>,
db: &Pool<Postgres>,
job: &QueuedJob,
schedule: &Schedule,
script_path: &str,
w_id: &str,
) -> windmill_common::error::Result<()> {
tracing::info!(
"Schedule {} scheduling next job for {} in {w_id}",
schedule.path,
schedule.script_path
);
if schedule.enabled && script_path == schedule.script_path {
let push_next_job_future = async {
let mut tx: QueueTransaction<'_, _> = (rsmq.clone(), db.begin().await?).into();
tx = push_scheduled_job(db, tx, &schedule).await?;
tx.commit().await?;
Ok::<(), Error>(())
};
match push_next_job_future.await {
Ok(_) => Ok(()),
Err(err) => {
match sqlx::query!(
"UPDATE schedule SET enabled = false, error = $1 WHERE workspace_id = $2 AND path = $3",
err.to_string(),
&schedule.workspace_id,
&schedule.path
)
.execute(db).await {
Ok(_) => {
match err {
Error::QuotaExceeded(_) => {}
_ => {
report_error_to_workspace_handler_or_critical_side_channel(rsmq, job, db,
format!("Could not schedule next job for {} with err {}. Schedule disabled", schedule.path, err)
).await;
}
}
Ok(())
}
Err(disable_err) => match err {
Error::QuotaExceeded(_) => Err(err),
_ => {
report_error_to_workspace_handler_or_critical_side_channel(rsmq, job, db,
format!("Could not schedule next job for {} and could not disable schedule with err {}. Will retry", schedule.path, disable_err)
).await;
Err(to_anyhow(disable_err).into())
}
},
}
}
}
} else {
if script_path != schedule.script_path {
tracing::warn!(
"Schedule {} in {w_id} has a different script path than the job. Not scheduling again", schedule.path
);
} else {
tracing::info!(
"Schedule {} in {w_id} is disabled. Not scheduling again.",
schedule.path
);
}
Ok(())
}
}
#[derive(Clone, Serialize, FromRow)]
struct CompletedJobSubset {
success: bool,
result: Option<sqlx::types::Json<Box<RawValue>>>,
started_at: chrono::DateTime<chrono::Utc>,
}
async fn apply_schedule_handlers<
'a,
'c,
T: Serialize + Send + Sync,
R: rsmq_async::RsmqConnection + Clone + Send + 'c,
>(
rsmq: Option<R>,
db: &Pool<Postgres>,
schedule: &Schedule,
script_path: &str,
w_id: &str,
success: bool,
result: Json<&'a T>,
job_id: Uuid,
started_at: DateTime<Utc>,
job_priority: Option<i16>,
) -> windmill_common::error::Result<()> {
if !success {
#[cfg(feature = "enterprise")]
if let Some(on_failure_path) = schedule.on_failure.clone() {
let times = schedule.on_failure_times.unwrap_or(1).max(1);
let exact = schedule.on_failure_exact.unwrap_or(false);
if times > 1 || exact {
let past_jobs = sqlx::query_as::<_, CompletedJobSubset>(
"SELECT success, result, started_at FROM completed_job WHERE workspace_id = $1 AND schedule_path = $2 AND script_path = $3 AND id != $4 ORDER BY created_at DESC LIMIT $5",
)
.bind(&schedule.workspace_id)
.bind(&schedule.path)
.bind(script_path)
.bind(job_id)
.bind(if exact { times } else { times - 1 } as i64,)
.fetch_all(db).await?;
let match_times = if exact {
past_jobs.len() == times as usize
&& past_jobs[..(times - 1) as usize].iter().all(|j| !j.success)
&& past_jobs[(times - 1) as usize].success
} else {
past_jobs.len() == ((times - 1) as usize)
&& past_jobs.iter().all(|j| !j.success)
};
if !match_times {
return Ok(());
}
}
push_error_handler(
db,
rsmq,
job_id,
Some(schedule.path.to_string()),
Some(script_path.to_string()),
schedule.is_flow,
w_id,
&on_failure_path,
result,
Some(times),
Some(started_at),
schedule.on_failure_extra_args.clone(),
&schedule.email,
true,
false,
job_priority,
)
.await?;
}
} else {
#[cfg(feature = "enterprise")]
if let Some(on_recovery_path) = schedule.on_recovery.clone() {
let mut tx: QueueTransaction<'_, R> = (rsmq.clone(), db.begin().await?).into();
let times = schedule.on_recovery_times.unwrap_or(1).max(1);
let past_jobs = sqlx::query_as::<_, CompletedJobSubset>(
"SELECT success, result, started_at FROM completed_job WHERE workspace_id = $1 AND schedule_path = $2 AND script_path = $3 AND id != $4 ORDER BY created_at DESC LIMIT $5",
)
.bind(&schedule.workspace_id)
.bind(&schedule.path)
.bind(script_path)
.bind(job_id)
.bind(times as i64)
.fetch_all(db).await?;
if past_jobs.len() < times as usize {
return Ok(());
}
let n_times_successful = past_jobs[..(times - 1) as usize].iter().all(|j| j.success);
if !n_times_successful {
return Ok(());
}
let failed_job = past_jobs[past_jobs.len() - 1].clone();
if !failed_job.success {
tx = handle_recovered_schedule(
db,
tx,
job_id,
&schedule.path,
script_path,
schedule.is_flow,
w_id,
&on_recovery_path,
failed_job,
result,
times,
started_at,
schedule.on_recovery_extra_args.clone(),
)
.await?;
}
tx.commit().await?;
}
}
Ok(())
}
pub async fn push_error_handler<
'a,
'c,
T: Serialize + Send + Sync,
R: rsmq_async::RsmqConnection + Clone + Send + 'c,
>(
db: &Pool<Postgres>,
rsmq: Option<R>,
job_id: Uuid,
schedule_path: Option<String>,
script_path: Option<String>,
is_flow: bool,
w_id: &str,
on_failure_path: &str,
result: Json<&'a T>,
failed_times: Option<i32>,
started_at: Option<DateTime<Utc>>,
extra_args: Option<Json<Box<RawValue>>>,
email: &str,
is_schedule_error_handler: bool,
is_global_error_handler: bool,
priority: Option<i16>,
) -> windmill_common::error::Result<Uuid> {
let handler_w_id = if is_global_error_handler {
"admins"
} else {
w_id
};
let (payload, tag) =
get_payload_tag_from_prefixed_path(on_failure_path, db, handler_w_id).await?;
let mut extra = HashMap::new();
if let Some(schedule_path) = schedule_path {
extra.insert("schedule_path".to_string(), to_raw_value(&schedule_path));
}
extra.insert("workspace_id".to_string(), to_raw_value(&w_id));
extra.insert("job_id".to_string(), to_raw_value(&job_id));
extra.insert("path".to_string(), to_raw_value(&script_path));
extra.insert("is_flow".to_string(), to_raw_value(&is_flow));
extra.insert("started_at".to_string(), to_raw_value(&started_at));
extra.insert("email".to_string(), to_raw_value(&email));
if let Some(failed_times) = failed_times {
extra.insert("failed_times".to_string(), to_raw_value(&failed_times));
}
if let Some(args_v) = extra_args {
if let Ok(args_m) = serde_json::from_str::<HashMap<String, Box<RawValue>>>(args_v.get()) {
extra.extend(args_m);
} else {
return Err(error::Error::ExecutionErr(
"args of scripts needs to be dict".to_string(),
));
}
}
// TODO(gbouv): REMOVE THIS after December 1st 2023 and ping users to re-save their error handlers
if on_failure_path
.to_string()
.eq("script/hub/5792/workspace-or-schedule-error-handler-slack")
{
// default slack error handler being used -> we need to inject the slack token
let slack_resource = format!("$res:{WORKSPACE_SLACK_BOT_TOKEN_PATH}");
extra.insert("slack".to_string(), to_raw_value(&slack_resource));
}
let result = sanitize_result(result);
let tx = PushIsolationLevel::IsolatedRoot(db.clone(), rsmq);
let (uuid, tx) = push(
&db,
tx,
handler_w_id,
payload,
PushArgs { extra, args: result },
if is_global_error_handler {
"global"
} else if is_schedule_error_handler {
SCHEDULE_ERROR_HANDLER_USERNAME
} else {
ERROR_HANDLER_USERNAME
},
if is_global_error_handler {
SUPERADMIN_SECRET_EMAIL
} else if is_schedule_error_handler {
SCHEDULE_ERROR_HANDLER_USER_EMAIL
} else {
ERROR_HANDLER_USER_EMAIL
},
if is_global_error_handler {
SUPERADMIN_SECRET_EMAIL.to_string()
} else {
ERROR_HANDLER_USER_GROUP.to_string()
},
None,
None,
Some(job_id),
Some(job_id),
None,
false,
false,
None,
true,
tag,
None,
None,
priority,
)
.await?;
tx.commit().await?;
return Ok(uuid);
}
fn sanitize_result<T: Serialize + Send + Sync>(result: Json<&T>) -> HashMap<String, Box<RawValue>> {
let as_str = serde_json::to_string(result.0).unwrap_or_else(|_| "{}".to_string());
serde_json::from_str::<HashMap<String, Box<RawValue>>>(&as_str)
.unwrap_or_else(|_| [("error".to_string(), RawValue::from_string(as_str).unwrap())].into())
}
// #[derive(Serialize)]
// pub struct RecoveryValue<T> {
// error_started_at: chrono::DateTime<Utc>,
// schedule_path: String,
// path: String,
// is_flow: boolean,
// extra_args: serde_json::Value
// }
async fn handle_recovered_schedule<
'a,
'c,
T: Serialize + Send + Sync,
R: rsmq_async::RsmqConnection + Clone + Send + 'c,
>(
db: &Pool<Postgres>,
tx: QueueTransaction<'c, R>,
job_id: Uuid,
schedule_path: &str,
script_path: &str,
is_flow: bool,
w_id: &str,
on_recovery_path: &str,
error_job: CompletedJobSubset,
successful_job_result: Json<&'a T>,
successful_times: i32,
successful_job_started_at: DateTime<Utc>,
extra_args: Option<Json<Box<RawValue>>>,
) -> windmill_common::error::Result<QueueTransaction<'c, R>> {
let (payload, tag) = get_payload_tag_from_prefixed_path(on_recovery_path, db, w_id).await?;
let mut extra = HashMap::new();
extra.insert(
"error_started_at".to_string(),
to_raw_value(&error_job.started_at),
);
extra.insert("schedule_path".to_string(), to_raw_value(&schedule_path));
extra.insert("path".to_string(), to_raw_value(&script_path));
extra.insert("is_flow".to_string(), to_raw_value(&is_flow));
extra.insert(
"success_result".to_string(),
serde_json::from_str::<Box<RawValue>>(
&serde_json::to_string(&successful_job_result).unwrap(),
)
.unwrap_or_else(|_| serde_json::value::RawValue::from_string("{}".to_string()).unwrap()),
);
extra.insert("success_times".to_string(), to_raw_value(&successful_times));
extra.insert(
"success_started_at".to_string(),
to_raw_value(&successful_job_started_at),
);
if let Some(args_v) = extra_args {
if let Ok(args_m) = serde_json::from_str::<HashMap<String, Box<RawValue>>>(args_v.get()) {
extra.extend(args_m);
} else {
return Err(error::Error::ExecutionErr(
"args of scripts needs to be dict".to_string(),
));
}
}
// TODO(gbouv): REMOVE THIS after December 1st 2023 and ping users to re-save their error handlers
if on_recovery_path
.to_string()
.eq("script/hub/2430/slack/schedule-recovery-handler-slack")
{
// default slack error handler being used -> we need to inject the slack token
let slack_resource = format!("$res:{WORKSPACE_SLACK_BOT_TOKEN_PATH}");
extra.insert("slack".to_string(), to_raw_value(&slack_resource));
}
let args = error_job
.result
.and_then(|x| serde_json::from_str::<HashMap<String, Box<RawValue>>>(x.0.get()).ok())
.unwrap_or_else(HashMap::new);
let tx = PushIsolationLevel::Transaction(tx);
let (uuid, tx) = push(
&db,
tx,
w_id,
payload,
PushArgs { extra: extra, args: args },
SCHEDULE_RECOVERY_HANDLER_USERNAME,
SCHEDULE_RECOVERY_HANDLER_USER_EMAIL,
ERROR_HANDLER_USER_GROUP.to_string(),
None,
None,
Some(job_id),
Some(job_id),
None,
false,
false,
None,
true,
tag,
None,
None,
None,
)
.await?;
tracing::info!(
"Pushed on_recovery job {} for {} to queue",
uuid,
schedule_path
);
return Ok(tx);
}
pub async fn pull<R: rsmq_async::RsmqConnection + Send + Clone>(
db: &Pool<Postgres>,
rsmq: Option<R>,
suspend_first: bool,
) -> windmill_common::error::Result<Option<QueuedJob>> {
loop {
let job = pull_single_job_and_mark_as_running_no_concurrency_limit(
db,
rsmq.clone(),
suspend_first,
)
.await?;
if job.is_none() {
return Ok(None);
}
let has_concurent_limit = job.as_ref().unwrap().concurrent_limit.is_some();
#[cfg(not(feature = "enterprise"))]
if has_concurent_limit {
tracing::error!("Concurrent limits are an EE feature only, ignoring constraints")
}
#[cfg(not(feature = "enterprise"))]
let has_concurent_limit = false;
// concurrency check. If more than X jobs for this path are already running, we re-queue and pull another job from the queue
let pulled_job = job.unwrap();
if pulled_job.script_path.is_none() || !has_concurent_limit || pulled_job.canceled {
#[cfg(feature = "prometheus")]
if METRICS_ENABLED.load(std::sync::atomic::Ordering::Relaxed) {
QUEUE_PULL_COUNT.inc();
}
return Ok(Option::Some(pulled_job));
}
let itx = db.begin().await?;
let mut tx: QueueTransaction<'_, _> = (rsmq.clone(), itx).into();
// Else the job is subject to concurrency limits
let job_script_path = pulled_job.script_path.clone().unwrap();
let job_concurrency_key = match concurrency_key(db, &pulled_job).await {
Ok(key) => key,
Err(e) => {
tracing::error!(
"Could not get concurrency key for job {} defaulting to default key: {e:?}",
pulled_job.id
);
legacy_concurrency_key(db, &pulled_job)
.await
.unwrap_or_else(|| pulled_job.full_path_with_workspace())
}
};
tracing::debug!("Concurrency key is '{}'", job_concurrency_key);
let job_custom_concurrent_limit = pulled_job.concurrent_limit.unwrap();
// setting concurrency_time_window to 0 will count only the currently running jobs
let job_custom_concurrency_time_window_s =
pulled_job.concurrency_time_window_s.unwrap_or(0);
tracing::debug!(
"Job concurrency limit is {} per {}s",
job_custom_concurrent_limit,
job_custom_concurrency_time_window_s
);
sqlx::query_scalar!(
"SELECT null FROM queue WHERE id = $1 FOR UPDATE",
pulled_job.id
)
.fetch_one(&mut tx)
.await
.context("lock job in queue")?;
let jobs_uuids_init_json_value = serde_json::from_str::<serde_json::Value>(
format!("{{\"{}\": {{}}}}", pulled_job.id.hyphenated().to_string()).as_str(),
)
.expect("Unable to serialize job_uuids column to proper JSON");
let running_job = sqlx::query_scalar!(
"INSERT INTO concurrency_counter(concurrency_id, job_uuids) VALUES ($1, $2)
ON CONFLICT (concurrency_id)
DO UPDATE SET job_uuids = jsonb_set(concurrency_counter.job_uuids, array[$3], '{}')
RETURNING (SELECT COUNT(*) FROM jsonb_object_keys(job_uuids))",
job_concurrency_key,
jobs_uuids_init_json_value,
pulled_job.id.hyphenated().to_string(),
)
.fetch_one(&mut tx)
.await
.map_err(|e| {
Error::InternalErr(format!(
"Error getting concurrency count for script path {job_script_path}: {e:#}"
))
})?;
tracing::debug!("running_job: {}", running_job.unwrap_or(0));
let completed_count = sqlx::query!(
"SELECT COUNT(*) as count, COALESCE(MAX(ended_at), now() - INTERVAL '1 second' * $2) as max_ended_at FROM concurrency_key WHERE key = $1 AND ended_at >= (now() - INTERVAL '1 second' * $2)",
job_concurrency_key,
f64::from(job_custom_concurrency_time_window_s),
).fetch_one(&mut tx).await.map_err(|e| {
Error::InternalErr(format!(
"Error getting completed count for key {job_concurrency_key}: {e:#}"
))
})?;
let min_started_at = sqlx::query!(
"SELECT COALESCE((SELECT MIN(started_at) as min_started_at
FROM queue
WHERE script_path = $1 AND job_kind != 'dependencies' AND running = true AND workspace_id = $2 AND canceled = false
GROUP BY script_path), $3) as min_started_at, now() AS now",
job_script_path,
&pulled_job.workspace_id,
completed_count.max_ended_at
)
.fetch_one(&mut tx)
.await
.map_err(|e| {
Error::InternalErr(format!(
"Error getting concurrency count for script path {job_script_path}: {e:#}"
))
})?;
let concurrent_jobs_for_this_script =
completed_count.count.unwrap_or_default() as i32 + running_job.unwrap_or(0) as i32;
tracing::debug!(
"Current concurrent jobs for this script: {}",
concurrent_jobs_for_this_script
);
if concurrent_jobs_for_this_script <= job_custom_concurrent_limit {
#[cfg(feature = "prometheus")]
if METRICS_ENABLED.load(std::sync::atomic::Ordering::Relaxed) {
QUEUE_PULL_COUNT.inc();
}
tx.commit().await?;
return Ok(Option::Some(pulled_job));
}
let x = sqlx::query_scalar!(
"UPDATE concurrency_counter SET job_uuids = job_uuids - $2 WHERE concurrency_id = $1 RETURNING (SELECT COUNT(*) FROM jsonb_object_keys(job_uuids))",
job_concurrency_key,
pulled_job.id.hyphenated().to_string(),
)
.fetch_one(&mut tx)
.await
.map_err(|e| {
Error::InternalErr(format!(
"Error decreasing concurrency count for script path {job_script_path}: {e:#}"
))
})?;
tracing::debug!("running_job after decrease: {}", x.unwrap_or(0));
let job_uuid: Uuid = pulled_job.id;
let avg_script_duration: Option<i64> = sqlx::query_scalar!(
"SELECT CAST(ROUND(AVG(duration_ms), 0) AS BIGINT) AS avg_duration_s FROM
(SELECT duration_ms FROM concurrency_key LEFT JOIN completed_job ON completed_job.id = concurrency_key.job_id WHERE key = $1 AND ended_at IS NOT NULL
ORDER BY ended_at
DESC LIMIT 10) AS t",
job_concurrency_key
)
.fetch_one(&mut tx)
.await?;
tracing::info!("avg script duration computed: {:?}", avg_script_duration);
// let before_me = sqlx::query!(
// "SELECT schedu FROM queue WHERE script_path = $1 AND job_kind != 'dependencies' AND running = true AND workspace_id = $2 AND canceled = false AND started_at < $3 ORDER BY started_at DESC LIMIT 1",
// job_script_path,
// &pulled_job.workspace_id,
// min_started_at.now.unwrap()
// )
// optimal scheduling is: 'older_job_in_concurrency_time_window_started_timestamp + script_avg_duration + concurrency_time_window_s'
let inc = Duration::try_milliseconds(
avg_script_duration.map(|x| i64::from(x + 100)).unwrap_or(0),
)
.unwrap_or_default()
.max(Duration::try_seconds(1).unwrap_or_default())
+ Duration::try_seconds(i64::from(job_custom_concurrency_time_window_s))
.unwrap_or_default();
let now = min_started_at.now.unwrap();
let min_started_p_inc = (min_started_at.min_started_at.unwrap_or(now) + inc)
.max(now + Duration::try_seconds(3).unwrap_or_default());
let mut estimated_next_schedule_timestamp = min_started_p_inc;
loop {
let nestimated = estimated_next_schedule_timestamp + inc;
let jobs_in_window = sqlx::query_scalar!(
"SELECT COUNT(*) FROM queue LEFT JOIN concurrency_key ON concurrency_key.job_id = queue.id
WHERE key = $1 AND running = false AND canceled = false AND scheduled_for >= $2 AND scheduled_for < $3",
job_concurrency_key,
estimated_next_schedule_timestamp,
nestimated
).fetch_optional(&mut tx).await?.flatten().unwrap_or(0) as i32;
tracing::info!("estimated_next_schedule_timestamp: {:?}, jobs_in_window: {jobs_in_window}, nestimated: {nestimated}, inc: {inc}", estimated_next_schedule_timestamp);
if jobs_in_window < job_custom_concurrent_limit {
break;
} else {
estimated_next_schedule_timestamp = nestimated;
}
}
tracing::info!("Job '{}' from path '{}' with concurrency key '{}' has reached its concurrency limit of {} jobs run in the last {} seconds. This job will be re-queued for next execution at {}",
job_uuid, job_script_path, job_concurrency_key, job_custom_concurrent_limit, job_custom_concurrency_time_window_s, estimated_next_schedule_timestamp);
let job_log_event = format!(
"\nRe-scheduled job to {estimated_next_schedule_timestamp} due to concurrency limits with key {job_concurrency_key} and limit {job_custom_concurrent_limit} in the last {job_custom_concurrency_time_window_s} seconds",
);
let _ = append_logs(job_uuid, pulled_job.workspace_id, job_log_event, db).await;
if rsmq.is_some() {
// if let Some(ref mut rsmq) = tx.rsmq {
// if using redis, only one message at a time can be poped from the queue. Process only this message and move to the next elligible job
// In this case, the job might be a job from the same script path, but we can't optimise this further
// if using posgtres, then we're able to re-queue the entire batch of scheduled job for this script_path, so we do it
let requeued_job_tag = sqlx::query_scalar::<_, String>(&format!(
"UPDATE queue
SET running = false
, started_at = null
, scheduled_for = '{estimated_next_schedule_timestamp}'
, last_ping = null
WHERE id = '{job_uuid}'
RETURNING tag"
))
.fetch_one(&mut tx)
.await
.map_err(|e| Error::InternalErr(format!("Could not update and re-queue job {job_uuid}. The job will be marked as running but it is not running: {e:#}")))?;
if let Some(ref mut rsmq) = tx.rsmq {
rsmq.send_message(
job_uuid.to_bytes_le().to_vec(),
Option::Some(estimated_next_schedule_timestamp),
requeued_job_tag,
);
}
tx.commit().await?;
} else {
// if using posgtres, then we're able to re-queue the entire batch of scheduled job for this script_path, so we do it
sqlx::query!(
"UPDATE queue
SET running = false
, started_at = null
, scheduled_for = $1
, last_ping = null
WHERE id = $2",
estimated_next_schedule_timestamp,
job_uuid,
)
.fetch_all(&mut tx)
.await
.map_err(|e| Error::InternalErr(format!("Could not update and re-queue job {job_uuid}. The job will be marked as running but it is not running: {e:#}")))?;
tx.commit().await?
}
}
}
async fn pull_single_job_and_mark_as_running_no_concurrency_limit<
'c,
R: rsmq_async::RsmqConnection + Send + Clone,
>(
db: &Pool<Postgres>,
rsmq: Option<R>,
suspend_first: bool,
) -> windmill_common::error::Result<Option<QueuedJob>> {
let job: Option<QueuedJob> = if let Some(mut rsmq) = rsmq {
#[cfg(feature = "benchmark")]
let instant = Instant::now();
// TODO: REDIS: Race conditions / replace last_ping
// TODO: shuffle this list to have fairness
let mut all_tags = WORKER_CONFIG.read().await.worker_tags.clone();
let mut msg: Option<_> = None;
let mut tag = None;
while msg.is_none() && !all_tags.is_empty() {
let ntag = all_tags.pop().unwrap();
tag = Some(ntag.clone());
msg = rsmq
.receive_message::<Vec<u8>>(&ntag, Some(10))
.await
.map_err(|e| anyhow::anyhow!(e))?;
}
// #[cfg(feature = "benchmark")]
// println!("rsmq 1: {:?}", instant.elapsed());
// println!("3.1: {:?} {rs}", instant.elapsed());
if let Some(msg) = msg {
let uuid = Uuid::from_bytes_le(
msg.message
.try_into()
.map_err(|_| anyhow::anyhow!("Failed to parsed Redis message"))?,
);
let m2r = sqlx::query(
"UPDATE queue
SET running = true
, started_at = coalesce(started_at, now())
, last_ping = now()
, suspend_until = null
WHERE id = $1
RETURNING id, workspace_id, parent_job, created_by, created_at, started_at, scheduled_for,
running, script_hash, script_path, args, right(logs, 900000) as logs, raw_code, canceled, canceled_by,
canceled_reason, last_ping, job_kind, env_id, schedule_path, permissioned_as,
flow_status, raw_flow, is_flow_step, language, suspend, suspend_until,
same_worker, raw_lock, pre_run_error, email, visible_to_owner, mem_peak,
root_job, leaf_jobs, tag, concurrent_limit, concurrency_time_window_s,
timeout, flow_step_id, cache_ttl, priority",
)
.bind(uuid)
.fetch_optional(db)
.await?;
let m2 = if let Some(row) = m2r {
Some(QueuedJob::from_row(&row)?)
} else {
None
};
rsmq.delete_message(&tag.unwrap(), &msg.id)
.await
.map_err(|e| anyhow::anyhow!(e))?;
#[cfg(feature = "benchmark")]
println!("rsmq 2: {:?}", instant.elapsed());
m2
} else {
None
}
} else {
/* Jobs can be started if they:
* - haven't been started before,
* running = false
* - are flows with a step that needed resume,
* suspend_until is non-null
* and suspend = 0 when the resume messages are received
* or suspend_until <= now() if it has timed out */
let config = WORKER_CONFIG.read().await.clone();
let tags = config.worker_tags.clone();
#[cfg(not(feature = "enterprise"))]
let priority_tags_sorted = vec![PriorityTags { priority: 0, tags: tags.clone() }];
#[cfg(feature = "enterprise")]
let priority_tags_sorted = config.priority_tags_sorted.clone();
drop(config);
let r = if suspend_first {
sqlx::query("UPDATE queue
SET running = true
, started_at = coalesce(started_at, now())
, last_ping = now()
, suspend_until = null
WHERE id = (
SELECT id
FROM queue
WHERE suspend_until IS NOT NULL AND (suspend <= 0 OR suspend_until <= now()) AND tag = ANY($1)
ORDER BY priority DESC NULLS LAST, created_at
FOR UPDATE SKIP LOCKED
LIMIT 1
)
RETURNING id, workspace_id, parent_job, created_by, created_at, started_at, scheduled_for,
running, script_hash, script_path, args, null as logs, raw_code, canceled, canceled_by,
canceled_reason, last_ping, job_kind, env_id, schedule_path, permissioned_as,
flow_status, raw_flow, is_flow_step, language, suspend, suspend_until,
same_worker, raw_lock, pre_run_error, email, visible_to_owner, mem_peak,
root_job, leaf_jobs, tag, concurrent_limit, concurrency_time_window_s,
timeout, flow_step_id, cache_ttl, priority")
.bind(tags)
.fetch_optional(db)
.await?
} else {
None
};
let r = if let Some(row) = r {
Some(QueuedJob::from_row(&row)?)
} else {
None
};
if r.is_none() {
// #[cfg(feature = "benchmark")]
// let instant = Instant::now();
let mut highest_priority_job: Option<QueuedJob> = None;
for priority_tags in priority_tags_sorted {
let r = sqlx::query(
"UPDATE queue
SET running = true
, started_at = coalesce(started_at, now())
, last_ping = now()
, suspend_until = null
WHERE id = (
SELECT id
FROM queue
WHERE running = false AND scheduled_for <= now() AND tag = ANY($1)
ORDER BY priority DESC NULLS LAST, scheduled_for, created_at
FOR UPDATE SKIP LOCKED
LIMIT 1
)
RETURNING id, workspace_id, parent_job, created_by, created_at, started_at, scheduled_for,
running, script_hash, script_path, args, null as logs, raw_code, canceled, canceled_by,
canceled_reason, last_ping, job_kind, env_id, schedule_path, permissioned_as,
flow_status, raw_flow, is_flow_step, language, suspend, suspend_until,
same_worker, raw_lock, pre_run_error, email, visible_to_owner, mem_peak,
root_job, leaf_jobs, tag, concurrent_limit, concurrency_time_window_s,
timeout, flow_step_id, cache_ttl, priority",
)
.bind(priority_tags.tags.clone())
.fetch_optional(db)
.await?;
if let Some(pulled_row) = r {
let pulled_job = QueuedJob::from_row(&pulled_row)?;
highest_priority_job = Some(pulled_job.clone());
tracing::debug!(
"Pulling for job {} with tags {:?} with priority {}",
pulled_job.id,
priority_tags.tags,
priority_tags.priority
);
break;
}
// else continue pulling for lower priority tags
}
// #[cfg(feature = "benchmark")]
// println!("pull query: {:?}", instant.elapsed());
highest_priority_job
} else {
r
}
};
Ok(job)
}
pub async fn custom_concurrency_key(
db: &Pool<Postgres>,
job_id: Uuid,
) -> Result<Option<String>, sqlx::Error> {
sqlx::query_scalar!("SELECT key FROM concurrency_key WHERE job_id = $1", job_id)
.fetch_optional(db) // this should no longer be fetch optional
.await
}
async fn legacy_concurrency_key(db: &Pool<Postgres>, queued_job: &QueuedJob) -> Option<String> {
let r = if queued_job.is_flow() {
sqlx::query_scalar!(
"SELECT value->>'concurrency_key' FROM flow WHERE path = $1 AND workspace_id = $2",
queued_job.script_path,
queued_job.workspace_id
)
.fetch_optional(db)
.await
} else {
sqlx::query_scalar!(
"SELECT concurrency_key FROM script WHERE hash = $1 AND workspace_id = $2",
queued_job.script_hash.unwrap_or(ScriptHash(0)).0,
queued_job.workspace_id
)
.fetch_optional(db)
.await
}
.ok()
.flatten()
.flatten();
r.map(|x| {
interpolate_args(
x,
&queued_job
.args
.clone()
.map(|x| x.0)
.unwrap_or_default()
.into(),
&queued_job.workspace_id,
)
})
}
async fn concurrency_key(
db: &Pool<Postgres>,
queued_job: &QueuedJob,
) -> windmill_common::error::Result<String> {
not_found_if_none(
custom_concurrency_key(db, queued_job.id).await?,
"ConcurrencyKey",
queued_job.id.to_string(),
)
}
fn interpolate_args(x: String, args: &PushArgs, workspace_id: &str) -> String {
// Save this value to avoid parsing twice
let workspaced = x.as_str().replace("$workspace", workspace_id).to_string();
if RE_ARG_TAG.is_match(&workspaced) {
let mut interpolated = workspaced.clone();
for cap in RE_ARG_TAG.captures_iter(&workspaced) {
let arg_name = cap.get(1).unwrap().as_str();
let arg_value = args
.args
.get(arg_name)
.or(args.extra.get(arg_name))
.map(|x| x.get())
.unwrap_or_default()
.trim_matches('"');
interpolated =
interpolated.replace(format!("$args[{}]", arg_name).as_str(), &arg_value);
}
interpolated
} else {
workspaced
}
}
fn fullpath_with_workspace(
workspace_id: &str,
script_path: Option<&String>,
job_kind: &JobKind,
) -> String {
let path = script_path.map(String::as_str).unwrap_or("tmp/main");
let is_flow = matches!(
job_kind,
&JobKind::Flow | &JobKind::FlowPreview | &JobKind::SingleScriptFlow
);
format!(
"{}/{}/{}",
workspace_id,
if is_flow { "flow" } else { "script" },
path,
)
}
#[derive(FromRow)]
pub struct ResultR {
result: Option<Json<Box<RawValue>>>,
}
#[derive(FromRow)]
pub struct ResultWithId {
result: Option<Json<Box<RawValue>>>,
id: Uuid,
}
pub async fn get_result_by_id(
db: Pool<Postgres>,
w_id: String,
flow_id: Uuid,
node_id: String,
json_path: Option<String>,
) -> error::Result<Box<RawValue>> {
match get_result_by_id_from_running_flow(
&db,
w_id.as_str(),
&flow_id,
node_id.as_str(),
json_path.clone(),
)
.await
{
Ok(res) => Ok(res),
Err(_) => {
let running_flow_job = sqlx::query_as::<_, QueuedJob>(
"SELECT * FROM queue WHERE COALESCE((SELECT root_job FROM queue WHERE id = $1), $1) = id AND workspace_id = $2"
).bind(flow_id)
.bind(&w_id)
.fetch_optional(&db)
.await?;
let restarted_from = windmill_common::utils::not_found_if_none(
running_flow_job
.map(|fj| fj.parse_flow_status())
.flatten()
.map(|status| status.restarted_from)
.flatten(),
"Flow result by id in leaf jobs",
format!("{}, {}", flow_id, node_id),
)?;
get_result_by_id_from_original_flow(
&db,
w_id.as_str(),
&restarted_from.flow_job_id,
node_id.as_str(),
json_path.clone(),
)
.await
}
}
}
#[async_recursion]
pub async fn get_result_by_id_from_running_flow(
db: &Pool<Postgres>,
w_id: &str,
flow_id: &Uuid,
node_id: &str,
json_path: Option<String>,
) -> error::Result<Box<RawValue>> {
let flow_job_result = sqlx::query!(
"SELECT leaf_jobs->$1::text as leaf_jobs, parent_job FROM queue WHERE COALESCE((SELECT root_job FROM queue WHERE id = $2), $2) = id AND workspace_id = $3",
node_id,
flow_id,
w_id,
)
.fetch_optional(db)
.await?;
let flow_job_result = windmill_common::utils::not_found_if_none(
flow_job_result,
"Flow result by id in leaf jobs",
format!("{}, {}", flow_id, node_id),
)?;
let job_result = flow_job_result
.leaf_jobs
.map(|x| serde_json::from_value(x).ok())
.flatten();
if job_result.is_none() && flow_job_result.parent_job.is_some() {
let parent_job = flow_job_result.parent_job.unwrap();
let root_job = sqlx::query_scalar!("SELECT root_job FROM queue WHERE id = $1", parent_job)
.fetch_optional(db)
.await?
.flatten()
.unwrap_or(parent_job);
return get_result_by_id_from_running_flow(db, w_id, &root_job, node_id, json_path).await;
}
let result_id = windmill_common::utils::not_found_if_none(
job_result,
"Flow result by id",
format!("{}, {}", flow_id, node_id),
)?;
extract_result_from_job_result(db, w_id, result_id, json_path).await
}
#[async_recursion]
async fn get_result_by_id_from_original_flow(
db: &Pool<Postgres>,
w_id: &str,
completed_flow_id: &Uuid,
node_id: &str,
json_path: Option<String>,
) -> error::Result<Box<RawValue>> {
let flow_job = sqlx::query_as::<_, CompletedJob>(
"SELECT *, null as labels FROM completed_job WHERE id = $1 AND workspace_id = $2",
)
.bind(completed_flow_id)
.bind(w_id)
.fetch_optional(db)
.await?;
let flow_job = windmill_common::utils::not_found_if_none(
flow_job,
"Flow result by id in leaf jobs",
format!("{}", completed_flow_id),
)?;
let mut leaf_jobs_for_flow = HashMap::<String, JobResult>::new();
compute_leaf_jobs_for_completed_flow(
db,
w_id,
flow_job.id,
Some(flow_job.clone()),
&mut leaf_jobs_for_flow,
)
.await?;
if !leaf_jobs_for_flow.contains_key(&node_id.to_string()) {
// if the flow is itself a restart flow, the step job might be from the upstream flow
let restarted_from = windmill_common::utils::not_found_if_none(
flow_job
.parse_flow_status()
.map(|status| status.restarted_from)
.flatten(),
"Flow result by id in leaf jobs",
format!("{}", completed_flow_id),
)?;
return get_result_by_id_from_original_flow(
db,
w_id,
&restarted_from.flow_job_id,
node_id,
json_path,
)
.await;
}
// if the job is in the leaf_jobs map, then fetch its result and return
let leaf_job_uuid = leaf_jobs_for_flow.get(&node_id.to_string()).unwrap();
extract_result_from_job_result(db, w_id, leaf_job_uuid.to_owned(), json_path).await
}
#[async_recursion]
async fn compute_leaf_jobs_for_completed_flow(
db: &Pool<Postgres>,
w_id: &str,
completed_flow_id: Uuid,
completed_flow_row: Option<CompletedJob>, // if provided, will be used, otherwise completed_flow_id must be set and the job definition will be pulled from DB
recursive_result: &mut HashMap<String, JobResult>,
) -> error::Result<()> {
let flow_status = match completed_flow_row {
Some(job) => job.parse_flow_status(),
None => {
let job_status_raw = sqlx::query_scalar!(
"SELECT flow_status FROM completed_job WHERE id = $1 AND workspace_id = $2",
completed_flow_id,
w_id,
)
.fetch_one(db)
.await?;
job_status_raw
.map(|raw| serde_json::from_value::<FlowStatus>(raw).ok())
.flatten()
}
};
let flow_job_status_modules = windmill_common::utils::not_found_if_none(
flow_status.map(|fs| fs.modules),
"Flow result by id in leaf jobs",
format!("{}", completed_flow_id),
)?;
let children_jobs = sqlx::query_as::<_, (Uuid, JobKind)>(
"SELECT id, job_kind FROM completed_job WHERE parent_job = $1 AND workspace_id = $2",
)
.bind(completed_flow_id)
.bind(w_id)
.fetch_all(db)
.await?;
for child_job in children_jobs {
let child_job_id = child_job.0;
let child_job_kind = child_job.1;
match child_job_kind {
JobKind::Script | JobKind::Preview | JobKind::Script_Hub => {
// if is potentially a leaf job. Get its step_id from the initial flow definition and add it to the result map
for module in &flow_job_status_modules {
if module.job().map(|id| id == child_job_id).unwrap_or(false) {
recursive_result.insert(module.id(), JobResult::SingleJob(child_job_id));
}
}
}
JobKind::Flow | JobKind::FlowPreview | JobKind::SingleScriptFlow => {
// Extract the leaf job for this flow and add them to the result map
for module in &flow_job_status_modules {
// we add the module as an element of ListJob for this step ID and recursiively extract leaf job of the sub-flow
match module {
FlowStatusModule::Success { flow_jobs: Some(jobs), .. } => {
let jobs_set: HashSet<Uuid> = HashSet::from_iter(jobs.iter().cloned());
if jobs_set.contains(&child_job_id) {
let new_list_job = match recursive_result.get(&module.id()) {
Some(JobResult::ListJob(jobs_list)) => {
let mut jobs_list_c = jobs_list.clone();
jobs_list_c.push(child_job_id);
jobs_list_c
}
_ => vec![child_job_id],
};
recursive_result
.insert(module.id(), JobResult::ListJob(new_list_job));
}
}
_ => {}
}
}
compute_leaf_jobs_for_completed_flow(
db,
w_id,
child_job_id,
None,
recursive_result,
)
.await?;
}
_ => {} // do nothing
}
}
Ok(())
}
async fn extract_result_from_job_result(
db: &Pool<Postgres>,
w_id: &str,
job_result: JobResult,
json_path: Option<String>,
) -> error::Result<Box<RawValue>> {
match job_result {
JobResult::ListJob(job_ids) => {
let rows = sqlx::query(
"SELECT id, result FROM completed_job WHERE id = ANY($1) AND workspace_id = $2",
)
.bind(job_ids.as_slice())
.bind(w_id)
.fetch_all(db)
.await?
.into_iter()
.filter_map(|x| {
ResultWithId::from_row(&x)
.ok()
.and_then(|x| x.result.map(|y| (x.id, y)))
})
.collect::<HashMap<Uuid, Json<Box<RawValue>>>>();
let result = job_ids
.into_iter()
.map(|id| {
rows.get(&id)
.map(|x| x.0.clone())
.unwrap_or_else(|| to_raw_value(&serde_json::Value::Null))
})
.collect::<Vec<_>>();
Ok(to_raw_value(&result))
}
JobResult::SingleJob(x) => Ok(sqlx::query(
"SELECT result #> $3 as result FROM completed_job WHERE id = $1 AND workspace_id = $2",
)
.bind(x)
.bind(w_id)
.bind(
json_path
.map(|x| x.split(".").map(|x| x.to_string()).collect::<Vec<_>>())
.unwrap_or_default(),
)
.fetch_optional(db)
.await?
.map(|r| {
ResultR::from_row(&r)
.ok()
.and_then(|x| x.result.map(|x| x.0))
})
.flatten()
.unwrap_or_else(|| to_raw_value(&serde_json::Value::Null))),
}
}
#[instrument(level = "trace", skip_all)]
pub async fn delete_job<'c, R: rsmq_async::RsmqConnection + Clone + Send>(
mut tx: QueueTransaction<'c, R>,
w_id: &str,
job_id: Uuid,
) -> windmill_common::error::Result<QueueTransaction<'c, R>> {
#[cfg(feature = "prometheus")]
if METRICS_ENABLED.load(std::sync::atomic::Ordering::Relaxed) {
QUEUE_DELETE_COUNT.inc();
}
let job_removed = sqlx::query_scalar!(
"DELETE FROM queue WHERE workspace_id = $1 AND id = $2 RETURNING 1",
w_id,
job_id
)
.fetch_optional(&mut tx)
.await;
if let Err(job_removed) = job_removed {
tracing::error!(
"Job {job_id} could not be deleted: {job_removed}. This is not necessarily an error, as the job might have been deleted by another process such as in the case of cancelling"
);
} else {
let job_removed = job_removed.unwrap().flatten().unwrap_or(0);
if job_removed != 1 {
tracing::error!("Job {job_id} could not be deleted, returned not 1: {job_removed}. This is not necessarily an error, as the job might have been deleted by another process such as in the case of cancelling");
}
}
tracing::debug!("Job {job_id} deleted");
Ok(tx)
}
pub async fn job_is_complete(db: &DB, id: Uuid, w_id: &str) -> error::Result<bool> {
Ok(sqlx::query_scalar!(
"SELECT EXISTS(SELECT 1 FROM completed_job WHERE id = $1 AND workspace_id = $2)",
id,
w_id
)
.fetch_one(db)
.await?
.unwrap_or(false))
}
pub async fn get_queued_job_tx<'c>(
id: Uuid,
w_id: &str,
tx: &mut Transaction<'c, Postgres>,
) -> error::Result<Option<QueuedJob>> {
let r = sqlx::query(
"SELECT *
FROM queue WHERE id = $1 AND workspace_id = $2",
)
.bind(id)
.bind(w_id)
.fetch_optional(&mut **tx)
.await?;
if let Some(row) = r {
Ok(Some(QueuedJob::from_row(&row)?.to_owned()))
} else {
Ok(None)
}
}
pub async fn get_queued_job(id: &Uuid, w_id: &str, db: &DB) -> error::Result<Option<QueuedJob>> {
let r = sqlx::query(
"SELECT *
FROM queue WHERE id = $1 AND workspace_id = $2",
)
.bind(id)
.bind(w_id)
.fetch_optional(db)
.await?;
if let Some(row) = r {
Ok(Some(QueuedJob::from_row(&row)?.to_owned()))
} else {
Ok(None)
}
}
pub enum PushIsolationLevel<'c, R: rsmq_async::RsmqConnection + Send + 'c> {
IsolatedRoot(DB, Option<R>),
Isolated(UserDB, Authed, Option<R>),
Transaction(QueueTransaction<'c, R>),
}
#[macro_export]
macro_rules! fetch_scalar_isolated {
( $query:expr, $tx:expr) => {
match $tx {
PushIsolationLevel::IsolatedRoot(db, rmsq) => {
let r = $query.fetch_optional(&db).await;
$tx = PushIsolationLevel::IsolatedRoot(db, rmsq);
r
}
PushIsolationLevel::Isolated(db, user, rsmq) => {
let mut ntx = db.clone().begin(&user).await?;
let r = $query.fetch_optional(&mut *ntx).await;
$tx = PushIsolationLevel::Isolated(db, user, rsmq);
r
}
PushIsolationLevel::Transaction(mut tx) => {
let r = $query.fetch_optional(&mut tx).await;
$tx = PushIsolationLevel::Transaction(tx);
r
}
}
};
}
use sqlx::types::JsonRawValue;
#[derive(Serialize, Debug)]
pub struct PushArgs {
#[serde(flatten)]
pub extra: HashMap<String, Box<RawValue>>,
#[serde(flatten)]
pub args: HashMap<String, Box<RawValue>>,
}
#[derive(Deserialize)]
pub struct DecodeQuery {
pub include_query: Option<String>,
}
#[derive(Deserialize)]
pub struct IncludeQuery {
pub include_query: Option<String>,
}
pub struct DecodeQueries(pub HashMap<String, Box<RawValue>>);
#[axum::async_trait]
impl<S> FromRequestParts<S> for DecodeQueries
where
S: Send + Sync,
{
type Rejection = Response;
async fn from_request_parts(parts: &mut Parts, _state: &S) -> Result<Self, Self::Rejection> {
Ok(DecodeQueries::from_uri(&parts.uri).unwrap_or_else(|| DecodeQueries(HashMap::new())))
}
}
impl DecodeQueries {
fn from_uri(uri: &Uri) -> Option<Self> {
let query = uri.query();
if query.is_none() {
return None;
}
let query = query.unwrap();
let include_query = serde_urlencoded::from_str::<IncludeQuery>(query)
.map(|x| x.include_query)
.ok()
.flatten()
.unwrap_or_default();
let parse_query_args = include_query
.split(",")
.map(|s| s.to_string())
.collect::<Vec<_>>();
let mut args = HashMap::new();
if !parse_query_args.is_empty() {
let queries =
serde_urlencoded::from_str::<HashMap<String, String>>(query).unwrap_or_default();
parse_query_args.iter().for_each(|h| {
if let Some(v) = queries.get(h) {
args.insert(h.to_string(), to_raw_value(v));
}
});
}
Some(DecodeQueries(args))
}
}
impl PushArgs {
pub fn insert<K: Into<String>, V: Into<Box<RawValue>>>(&mut self, k: K, v: V) {
self.extra.insert(k.into(), v.into());
}
}
#[derive(Deserialize)]
pub struct RequestQuery {
pub raw: Option<bool>,
pub include_header: Option<String>,
}
#[axum::async_trait]
impl<S> FromRequest<S, axum::body::Body> for PushArgs
where
S: Send + Sync,
{
type Rejection = Response;
async fn from_request(
req: Request<axum::body::Body>,
_state: &S,
) -> Result<Self, Self::Rejection> {
let (content_type, mut extra, use_raw) = {
let headers_map = req.headers();
let content_type_header = headers_map.get(CONTENT_TYPE);
let content_type = content_type_header.and_then(|value| value.to_str().ok());
let uri = req.uri();
let query = Query::<RequestQuery>::try_from_uri(uri).unwrap().0;
let mut extra = build_extra(&headers_map, query.include_header);
let query_decode = DecodeQueries::from_uri(uri);
if let Some(DecodeQueries(queries)) = query_decode {
extra.extend(queries);
}
let raw = query.raw.as_ref().is_some_and(|x| *x);
(content_type, extra, raw)
};
if content_type.is_none() || content_type.unwrap().starts_with("application/json") {
let bytes = Bytes::from_request(req, _state)
.await
.map_err(IntoResponse::into_response)?;
let str = String::from_utf8(bytes.to_vec())
.map_err(|e| Error::BadRequest(format!("invalid utf8: {}", e)).into_response())?;
if use_raw {
extra.insert("raw_string".to_string(), to_raw_value(&str));
}
let wrap_body = str.len() > 0 && str.chars().next().unwrap() != '{';
if wrap_body {
let args = serde_json::from_str::<Option<Box<RawValue>>>(&str)
.map_err(|e| Error::BadRequest(format!("invalid json: {}", e)).into_response())?
.unwrap_or_else(|| to_raw_value(&serde_json::Value::Null));
let mut hm = HashMap::new();
hm.insert("body".to_string(), args);
Ok(PushArgs { extra, args: hm })
} else {
let hm = serde_json::from_str::<Option<HashMap<String, Box<JsonRawValue>>>>(&str)
.map_err(|e| Error::BadRequest(format!("invalid json: {}", e)).into_response())?
.unwrap_or_else(HashMap::new);
Ok(PushArgs { extra, args: hm })
}
} else if content_type
.unwrap()
.starts_with("application/x-www-form-urlencoded")
{
let bytes = Bytes::from_request(req, _state)
.await
.map_err(IntoResponse::into_response)?;
if use_raw {
let raw_string = String::from_utf8(bytes.to_vec()).map_err(|e| {
Error::BadRequest(format!("invalid utf8: {}", e)).into_response()
})?;
extra.insert("raw_string".to_string(), to_raw_value(&raw_string));
}
let payload: HashMap<String, Option<String>> = serde_urlencoded::from_bytes(&bytes)
.map_err(|e| {
Error::BadRequest(format!("invalid urlencoded data: {}", e)).into_response()
})?;
let payload = payload
.into_iter()
.map(|(k, v)| (k, to_raw_value(&v)))
.collect::<HashMap<_, _>>();
return Ok(PushArgs { extra, args: payload });
} else {
Err(StatusCode::UNSUPPORTED_MEDIA_TYPE.into_response())
}
}
}
lazy_static::lazy_static! {
static ref INCLUDE_HEADERS: Vec<String> = std::env::var("INCLUDE_HEADERS")
.ok().map(|x| x
.split(',')
.map(|s| s.to_string())
.collect()).unwrap_or_default();
}
pub fn build_extra(
headers: &HeaderMap,
include_header: Option<String>,
) -> HashMap<String, Box<RawValue>> {
let mut args = HashMap::new();
let whitelist = include_header
.map(|s| s.split(",").map(|s| s.to_string()).collect::<Vec<_>>())
.unwrap_or_default();
whitelist
.iter()
.chain(INCLUDE_HEADERS.iter())
.for_each(|h| {
if let Some(v) = headers.get(h) {
args.insert(
h.to_string().to_lowercase().replace('-', "_"),
to_raw_value(&v.to_str().unwrap().to_string()),
);
}
});
args
}
impl PushArgs {
pub fn empty() -> Self {
PushArgs { extra: HashMap::new(), args: HashMap::new() }
}
}
pub fn empty_result() -> Box<RawValue> {
return JsonRawValue::from_string("{}".to_string()).unwrap();
}
impl From<HashMap<String, Box<JsonRawValue>>> for PushArgs {
fn from(value: HashMap<String, Box<JsonRawValue>>) -> Self {
PushArgs { extra: HashMap::new(), args: value }
}
}
// impl<T> From<PushArgsInner<T>> for PushArgs<T> {
// fn from(value: PushArgsInner<T>) -> Self {
// PushArgs::Unwrapped(value)
// }
// }
lazy_static::lazy_static! {
pub static ref RE_ARG_TAG: Regex = Regex::new(r#"\$args\[(\w+)\]"#).unwrap();
}
// #[instrument(level = "trace", skip_all)]
pub async fn push<'c, R: rsmq_async::RsmqConnection + Send + 'c>(
_db: &Pool<Postgres>,
mut tx: PushIsolationLevel<'c, R>,
workspace_id: &str,
job_payload: JobPayload,
args: PushArgs,
user: &str,
mut email: &str,
mut permissioned_as: String,
scheduled_for_o: Option<chrono::DateTime<chrono::Utc>>,
schedule_path: Option<String>,
parent_job: Option<Uuid>,
root_job: Option<Uuid>,
job_id: Option<Uuid>,
is_flow_step: bool,
mut same_worker: bool,
pre_run_error: Option<&windmill_common::error::Error>,
visible_to_owner: bool,
mut tag: Option<String>,
custom_timeout: Option<i32>,
flow_step_id: Option<String>,
_priority_override: Option<i16>,
) -> Result<(Uuid, QueueTransaction<'c, R>), Error> {
#[cfg(feature = "cloud")]
if *CLOUD_HOSTED {
let premium_workspace =
sqlx::query_scalar!("SELECT premium FROM workspace WHERE id = $1", workspace_id)
.fetch_one(_db)
.await
.map_err(|e| {
Error::InternalErr(format!(
"fetching if {workspace_id} is premium and overquota: {e:#}"
))
})?;
// we track only non flow steps
let (workspace_usage, user_usage) = if !matches!(
job_payload,
JobPayload::Flow { .. } | JobPayload::RawFlow { .. }
) {
let workspace_usage = sqlx::query_scalar!(
"INSERT INTO usage (id, is_workspace, month_, usage)
VALUES ($1, TRUE, EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date), 1)
ON CONFLICT (id, is_workspace, month_) DO UPDATE SET usage = usage.usage + 1
RETURNING usage.usage",
workspace_id
)
.fetch_one(_db)
.await
.map_err(|e| Error::InternalErr(format!("updating usage: {e:#}")))?;
let user_usage = if !premium_workspace {
Some(sqlx::query_scalar!(
"INSERT INTO usage (id, is_workspace, month_, usage)
VALUES ($1, FALSE, EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date), 1)
ON CONFLICT (id, is_workspace, month_) DO UPDATE SET usage = usage.usage + 1
RETURNING usage.usage",
email
)
.fetch_one(_db)
.await
.map_err(|e| Error::InternalErr(format!("updating usage: {e:#}")))?)
} else {
None
};
(Some(workspace_usage), user_usage)
} else {
(None, None)
};
if !premium_workspace {
let is_super_admin =
sqlx::query_scalar!("SELECT super_admin FROM password WHERE email = $1", email)
.fetch_optional(_db)
.await?
.unwrap_or(false);
if !is_super_admin {
if email != ERROR_HANDLER_USER_EMAIL
&& email != SCHEDULE_ERROR_HANDLER_USER_EMAIL
&& email != SCHEDULE_RECOVERY_HANDLER_USER_EMAIL
&& email != "worker@windmill.dev"
&& email != SUPERADMIN_SECRET_EMAIL
&& email != SUPERADMIN_SYNC_EMAIL
&& email != SUPERADMIN_NOTIFICATION_EMAIL
{
let user_usage = if let Some(user_usage) = user_usage {
user_usage
} else {
sqlx::query_scalar!(
"SELECT usage.usage + 1 FROM usage
WHERE is_workspace IS FALSE AND
month_ = EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date)
AND id = $1",
email
)
.fetch_optional(_db)
.await?
.flatten()
.unwrap_or(1)
};
if user_usage > MAX_FREE_EXECS
&& !matches!(job_payload, JobPayload::Dependencies { .. })
&& !matches!(job_payload, JobPayload::FlowDependencies { .. })
&& !matches!(job_payload, JobPayload::AppDependencies { .. })
{
return Err(error::Error::QuotaExceeded(format!(
"User {email} has exceeded the free usage limit of {MAX_FREE_EXECS} that applies outside of premium workspaces."
)));
}
let in_queue =
sqlx::query_scalar!("SELECT COUNT(id) FROM queue WHERE email = $1", email)
.fetch_one(_db)
.await?
.unwrap_or(0);
if in_queue > MAX_FREE_EXECS.into() {
return Err(error::Error::QuotaExceeded(format!(
"User {email} has exceeded the jobs in queue limit of {MAX_FREE_EXECS} that applies outside of premium workspaces."
)));
}
let concurrent_runs = sqlx::query_scalar!(
"SELECT COUNT(id) FROM queue WHERE running = true AND email = $1",
email
)
.fetch_one(_db)
.await?
.unwrap_or(0);
if concurrent_runs > MAX_FREE_CONCURRENT_RUNS.into() {
return Err(error::Error::QuotaExceeded(format!(
"User {email} has exceeded the concurrent runs limit of {MAX_FREE_CONCURRENT_RUNS} that applies outside of premium workspaces."
)));
}
}
if workspace_id != "demo" {
let workspace_usage = if let Some(workspace_usage) = workspace_usage {
workspace_usage
} else {
sqlx::query_scalar!(
"SELECT usage.usage + 1 FROM usage
WHERE is_workspace IS TRUE AND
month_ = EXTRACT(YEAR FROM current_date) * 12 + EXTRACT(MONTH FROM current_date)
AND id = $1",
workspace_id
)
.fetch_optional(_db)
.await?
.flatten()
.unwrap_or(1)
};
if workspace_usage > MAX_FREE_EXECS
&& !matches!(job_payload, JobPayload::Dependencies { .. })
&& !matches!(job_payload, JobPayload::FlowDependencies { .. })
&& !matches!(job_payload, JobPayload::AppDependencies { .. })
{
return Err(error::Error::QuotaExceeded(format!(
"Workspace {workspace_id} has exceeded the free usage limit of {MAX_FREE_EXECS} that applies outside of premium workspaces."
)));
}
let in_queue_workspace = sqlx::query_scalar!(
"SELECT COUNT(id) FROM queue WHERE workspace_id = $1",
workspace_id
)
.fetch_one(_db)
.await?
.unwrap_or(0);
if in_queue_workspace > MAX_FREE_EXECS.into() {
return Err(error::Error::QuotaExceeded(format!(
"Workspace {workspace_id} has exceeded the jobs in queue limit of {MAX_FREE_EXECS} that applies outside of premium workspaces."
)));
}
let concurrent_runs_workspace = sqlx::query_scalar!(
"SELECT COUNT(id) FROM queue WHERE running = true AND workspace_id = $1",
workspace_id
)
.fetch_one(_db)
.await?
.unwrap_or(0);
if concurrent_runs_workspace > MAX_FREE_CONCURRENT_RUNS.into() {
return Err(error::Error::QuotaExceeded(format!(
"Workspace {workspace_id} has exceeded the concurrent runs limit of {MAX_FREE_CONCURRENT_RUNS} that applies outside of premium workspaces."
)));
}
}
}
}
}
let (
script_hash,
script_path,
raw_code_tuple,
job_kind,
raw_flow,
flow_status,
language,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
dedicated_worker,
_low_level_priority,
) = match job_payload {
JobPayload::ScriptHash {
hash,
path,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
language,
dedicated_worker,
priority,
} => (
Some(hash.0),
Some(path),
None,
JobKind::Script,
None,
None,
Some(language),
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
dedicated_worker,
priority,
),
JobPayload::ScriptHub { path } => {
if path == "hub/7771/slack" || path == "hub/7836/slack" {
permissioned_as = SUPERADMIN_NOTIFICATION_EMAIL.to_string();
email = SUPERADMIN_NOTIFICATION_EMAIL;
}
let hub_script =
get_full_hub_script_by_path(StripPath(path.clone()), &HTTP_CLIENT, _db).await?;
(
None,
Some(path),
None,
// Some((script.content, script.lockfile)),
JobKind::Script_Hub,
None,
None,
Some(hub_script.language),
None,
None,
None,
None,
None,
None,
)
}
JobPayload::Code(RawCode {
content,
path,
hash,
language,
lock,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
dedicated_worker,
}) => (
hash,
path,
Some((content, lock)),
JobKind::Preview,
None,
None,
Some(language),
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
dedicated_worker,
None,
),
JobPayload::Dependencies { hash, language, path, dedicated_worker } => (
Some(hash.0),
Some(path),
None,
JobKind::Dependencies,
None,
None,
Some(language),
None,
None,
None,
None,
dedicated_worker,
None,
),
JobPayload::RawScriptDependencies { script_path, content, language } => (
None,
Some(script_path),
Some((content, None)),
JobKind::Dependencies,
None,
None,
Some(language),
None,
None,
None,
None,
None,
None,
),
JobPayload::FlowDependencies { path, dedicated_worker } => {
let value_json = fetch_scalar_isolated!(
sqlx::query_scalar!(
"SELECT value FROM flow WHERE path = $1 AND workspace_id = $2",
path,
workspace_id
),
tx
)?
.ok_or_else(|| Error::InternalErr(format!("not found flow at path {:?}", path)))?;
let value = serde_json::from_value::<FlowValue>(value_json).map_err(|err| {
Error::InternalErr(format!(
"could not convert json to flow for {path}: {err:?}"
))
})?;
(
None,
Some(path),
None,
JobKind::FlowDependencies,
Some(value.clone()),
Some(FlowStatus::new(&value)), // this is a new flow being pushed, flow_status is set to flow_value
None,
None,
None,
None,
None,
dedicated_worker,
None,
)
}
JobPayload::AppDependencies { path, version } => (
Some(version),
Some(path),
None,
JobKind::AppDependencies,
None,
None,
None,
None,
None,
None,
None,
None,
None,
),
JobPayload::RawFlow { mut value, path, restarted_from } => {
add_virtual_items_if_necessary(&mut value.modules);
let flow_status: FlowStatus = match restarted_from {
Some(restarted_from_val) => {
let (_, _, step_n, truncated_modules, _, user_states, cleanup_module) =
restarted_flows_resolution(
_db,
workspace_id,
Some(value.clone()),
restarted_from_val.flow_job_id,
restarted_from_val.step_id.as_str(),
restarted_from_val.branch_or_iteration_n,
)
.await?;
FlowStatus {
step: step_n,
modules: truncated_modules,
// failure_module is reset
failure_module: Box::new(FlowStatusModuleWParent {
parent_module: None,
module_status: FlowStatusModule::WaitingForPriorSteps {
id: "failure".to_string(),
},
}),
cleanup_module,
// retry status is reset
retry: RetryStatus { fail_count: 0, failed_jobs: vec![] },
// TODO: for now, flows with approval conditions aren't supported for restart
approval_conditions: None,
restarted_from: Some(RestartedFrom {
flow_job_id: restarted_from_val.flow_job_id,
step_id: restarted_from_val.step_id,
branch_or_iteration_n: restarted_from_val.branch_or_iteration_n,
}),
user_states,
}
}
_ => FlowStatus::new(&value), // this is a new flow being pushed, flow_status is set to flow_value
};
(
None,
path,
None,
JobKind::FlowPreview,
Some(value.clone()),
Some(flow_status),
None,
value.concurrency_key.clone(),
value.concurrent_limit.clone(),
value.concurrency_time_window_s,
value.cache_ttl.map(|x| x as i32),
None,
value.priority,
)
}
JobPayload::SingleScriptFlow {
path,
hash,
retry,
args,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
priority,
tag_override,
} => {
let mut input_transforms = HashMap::<String, InputTransform>::new();
for (arg_name, arg_value) in args {
input_transforms.insert(arg_name, InputTransform::Static { value: arg_value });
}
let flow_value = FlowValue {
modules: vec![FlowModule {
id: "a".to_string(),
value: windmill_common::worker::to_raw_value(
&windmill_common::flows::FlowModuleValue::Script {
input_transforms: input_transforms,
path: path.clone(),
hash: Some(hash),
tag_override: tag_override,
},
),
stop_after_if: None,
summary: None,
suspend: None,
mock: None,
retry: Some(retry),
sleep: None,
cache_ttl: None,
timeout: None,
priority: None,
delete_after_use: None,
continue_on_error: None,
}],
same_worker: false,
failure_module: None,
concurrency_time_window_s: concurrency_time_window_s,
concurrent_limit: concurrent_limit,
skip_expr: None,
cache_ttl: cache_ttl.map(|val| val as u32),
early_return: None,
concurrency_key: custom_concurrency_key.clone(),
priority: priority,
};
(
None,
Some(path),
None,
JobKind::Flow,
Some(flow_value.clone()),
Some(FlowStatus::new(&flow_value)), // this is a new flow being pushed, flow_status is set to flow_value
None,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
None,
priority,
)
}
JobPayload::Flow { path, dedicated_worker } => {
let value_json = fetch_scalar_isolated!(
sqlx::query_scalar!(
"SELECT value FROM flow WHERE path = $1 AND workspace_id = $2",
path,
workspace_id
),
tx
)?
.ok_or_else(|| Error::InternalErr(format!("not found flow at path {:?}", path)))?;
let mut value = serde_json::from_value::<FlowValue>(value_json).map_err(|err| {
Error::InternalErr(format!(
"could not convert json to flow for {path}: {err:?}"
))
})?;
let priority = value.priority;
add_virtual_items_if_necessary(&mut value.modules);
let cache_ttl = value.cache_ttl.map(|x| x as i32).clone();
let custom_concurrency_key = value.concurrency_key.clone();
let concurrency_time_window_s = value.concurrency_time_window_s.clone();
let concurrent_limit = value.concurrent_limit.clone();
let status = Some(FlowStatus::new(&value));
(
None,
Some(path),
None,
JobKind::Flow,
Some(value),
status, // this is a new flow being pushed, flow_status is set to flow_value
None,
custom_concurrency_key,
concurrent_limit,
concurrency_time_window_s,
cache_ttl,
dedicated_worker,
priority,
)
}
JobPayload::RestartedFlow { completed_job_id, step_id, branch_or_iteration_n } => {
let (
flow_path,
raw_flow,
step_n,
truncated_modules,
priority,
user_states,
cleanup_module,
) = restarted_flows_resolution(
_db,
workspace_id,
None,
completed_job_id,
step_id.as_str(),
branch_or_iteration_n,
)
.await?;
let restarted_flow_status = FlowStatus {
step: step_n,
modules: truncated_modules,
// failure_module is reset
failure_module: Box::new(FlowStatusModuleWParent {
parent_module: None,
module_status: FlowStatusModule::WaitingForPriorSteps {
id: "failure".to_string(),
},
}),
cleanup_module,
// retry status is reset
retry: RetryStatus { fail_count: 0, failed_jobs: vec![] },
// TODO: for now, flows with approval conditions aren't supported for restart
approval_conditions: None,
restarted_from: Some(RestartedFrom {
flow_job_id: completed_job_id,
step_id,
branch_or_iteration_n,
}),
user_states,
};
(
None,
flow_path,
None,
JobKind::Flow,
Some(raw_flow.clone()),
Some(restarted_flow_status),
None,
raw_flow.concurrency_key,
raw_flow.concurrent_limit,
raw_flow.concurrency_time_window_s,
raw_flow.cache_ttl.map(|x| x as i32),
None,
priority,
)
}
JobPayload::DeploymentCallback { path } => (
None,
Some(path),
None,
JobKind::DeploymentCallback,
None,
None,
None,
None,
None,
None,
None,
None,
None,
),
JobPayload::Identity => (
None,
None,
None,
JobKind::Identity,
None,
None,
None,
None,
None,
None,
None,
None,
None,
),
JobPayload::Noop => (
None,
None,
None,
JobKind::Noop,
None,
None,
None,
None,
None,
None,
None,
None,
None,
),
};
let final_priority: Option<i16>;
#[cfg(not(feature = "enterprise"))]
{
// priority is only available on EE. Do not compute it on CE
final_priority = None;
}
#[cfg(feature = "enterprise")]
{
final_priority = if *CLOUD_HOSTED {
// for cloud hosted instance, priority queues is disabled
None
} else if _priority_override.is_some() {
_priority_override
} else {
// else it takes the priority defined at the script/flow level, if it's a script or flow
_low_level_priority
}; // else it remains empty, i.e. no priority
}
let is_running = same_worker;
if let Some(flow) = raw_flow.as_ref() {
same_worker = same_worker || flow.same_worker;
for module in flow.modules.iter() {
if let Some(retry) = &module.retry {
if retry.max_attempts() > MAX_RETRY_ATTEMPTS {
Err(Error::BadRequest(format!(
"retry attempts exceeds the maximum of {MAX_RETRY_ATTEMPTS}"
)))?
}
if matches!(retry.max_interval(), Some(interval) if interval > MAX_RETRY_INTERVAL) {
let max = MAX_RETRY_INTERVAL.as_secs();
Err(Error::BadRequest(format!(
"retry interval exceeds the maximum of {max} seconds"
)))?
}
}
}
}
let (raw_code, raw_lock) = raw_code_tuple
.map(|e| (Some(e.0), e.1))
.unwrap_or_else(|| (None, None));
let per_workspace: bool = DEFAULT_TAGS_PER_WORKSPACE.load(std::sync::atomic::Ordering::Relaxed);
let tag = if dedicated_worker.is_some_and(|x| x) {
format!(
"{}:{}{}",
workspace_id,
if job_kind == JobKind::Flow || job_kind == JobKind::FlowDependencies {
"flow/"
} else {
""
},
script_path.clone().expect("dedicated script has a path")
)
} else {
if tag == Some("".to_string()) {
tag = None;
}
let interpolated_tag = tag.map(|x| interpolate_args(x, &args, workspace_id));
let default = || {
let ntag = if job_kind == JobKind::Flow
|| job_kind == JobKind::FlowPreview
|| job_kind == JobKind::SingleScriptFlow
|| job_kind == JobKind::Identity
{
"flow".to_string()
} else if job_kind == JobKind::Dependencies
|| job_kind == JobKind::FlowDependencies
|| job_kind == JobKind::DeploymentCallback
{
// using the dependency tag for deployment callback for now. We can create a separate tag when we need
"dependency".to_string()
} else {
"deno".to_string()
};
if per_workspace {
format!("{}-{}", ntag, workspace_id)
} else {
ntag
}
};
interpolated_tag.unwrap_or_else(|| {
language
.as_ref()
.map(|x| {
if per_workspace {
format!("{}-{}", x.as_str(), workspace_id)
} else {
x.as_str().to_string()
}
})
.unwrap_or_else(default)
})
};
let mut tx = match tx {
PushIsolationLevel::Isolated(user_db, authed, rsmq) => {
(rsmq, user_db.begin(&authed).await?).into()
}
PushIsolationLevel::IsolatedRoot(db, rsmq) => (rsmq, db.begin().await?).into(),
PushIsolationLevel::Transaction(tx) => tx,
};
let job_id: Uuid = if let Some(job_id) = job_id {
let conflicting_id = sqlx::query_scalar!(
"SELECT 1 FROM queue WHERE id = $1 UNION ALL select 1 FROM completed_job WHERE id = $1",
job_id
)
.fetch_optional(&mut tx)
.await?;
if conflicting_id.is_some() {
return Err(Error::BadRequest(format!(
"Job with id {job_id} already exists"
)));
}
job_id
} else {
Ulid::new().into()
};
if concurrent_limit.is_some() {
let concurrency_key = custom_concurrency_key
.map(|x| interpolate_args(x, &args, workspace_id))
.unwrap_or(fullpath_with_workspace(
workspace_id,
script_path.as_ref(),
&job_kind,
));
sqlx::query!(
"INSERT INTO concurrency_key(key, job_id) VALUES ($1, $2)",
concurrency_key,
job_id,
)
.execute(&mut tx)
.await
.map_err(|e| Error::InternalErr(format!("Could not insert concurrency_key={concurrency_key} for job_id={job_id} script_path={script_path:?} workspace_id={workspace_id}: {e:#}")))?;
}
let uuid = sqlx::query_scalar!(
"INSERT INTO queue
(workspace_id, id, running, parent_job, created_by, permissioned_as, scheduled_for,
script_hash, script_path, raw_code, raw_lock, args, job_kind, schedule_path, raw_flow, \
flow_status, is_flow_step, language, started_at, same_worker, pre_run_error, email, \
visible_to_owner, root_job, tag, concurrent_limit, concurrency_time_window_s, timeout, \
flow_step_id, cache_ttl, priority)
VALUES ($1, $2, $3, $4, $5, $6, COALESCE($7, now()), $8, $9, $10, $11, $12, $13, $14, $15, $16, $17, $18, CASE WHEN $3 THEN now() END, $19, $20, $21, $22, $23, $24, $25, $26, $27, $28, $29, $30) \
RETURNING id",
workspace_id,
job_id,
is_running,
parent_job,
user,
permissioned_as,
scheduled_for_o,
script_hash,
script_path.clone(),
raw_code,
raw_lock,
Json(args) as Json<PushArgs>,
job_kind.clone() as JobKind,
schedule_path,
raw_flow.map(|f| serde_json::json!(f)),
flow_status.map(|f| serde_json::json!(f)),
is_flow_step,
language as Option<ScriptLang>,
same_worker,
pre_run_error.map(|e| e.to_string()),
email,
visible_to_owner,
root_job,
tag,
concurrent_limit,
if concurrent_limit.is_some() { concurrency_time_window_s } else { None },
custom_timeout,
flow_step_id,
cache_ttl,
final_priority,
)
.fetch_one(&mut tx)
.await
.map_err(|e| Error::InternalErr(format!("Could not insert into queue {job_id} with tag {tag}, schedule_path {schedule_path:?}, script_path: {script_path:?}, email {email}, workspace_id {workspace_id}: {e:#}")))?;
// TODO: technically the job isn't queued yet, as the transaction can be rolled back. Should be solved when moving these metrics to the queue abstraction.
#[cfg(feature = "prometheus")]
if METRICS_ENABLED.load(std::sync::atomic::Ordering::Relaxed) {
QUEUE_PUSH_COUNT.inc();
}
{
let uuid_string = job_id.to_string();
let uuid_str = uuid_string.as_str();
let mut hm = HashMap::from([("uuid", uuid_str), ("permissioned_as", &permissioned_as)]);
let s: String;
let operation_name = match job_kind {
JobKind::Preview => "jobs.run.preview",
JobKind::Script => {
s = ScriptHash(script_hash.unwrap()).to_string();
hm.insert("hash", s.as_str());
"jobs.run.script"
}
JobKind::Flow => "jobs.run.flow",
JobKind::FlowPreview => "jobs.run.flow_preview",
JobKind::SingleScriptFlow => "jobs.run.single_script_flow",
JobKind::Script_Hub => "jobs.run.script_hub",
JobKind::Dependencies => "jobs.run.dependencies",
JobKind::Identity => "jobs.run.identity",
JobKind::Noop => "jobs.run.noop",
JobKind::FlowDependencies => "jobs.run.flow_dependencies",
JobKind::AppDependencies => "jobs.run.app_dependencies",
JobKind::DeploymentCallback => "jobs.run.deployment_callback",
};
let audit_author = if format!("u/{user}") != permissioned_as && user != permissioned_as {
AuditAuthor {
email: email.to_string(),
username: permissioned_as.trim_start_matches("u/").to_string(),
username_override: Some(user.to_string()),
}
} else {
AuditAuthor {
email: email.to_string(),
username: user.to_string(),
username_override: None,
}
};
audit_log(
&mut tx,
&audit_author,
operation_name,
ActionKind::Execute,
workspace_id,
script_path.as_ref().map(|x| x.as_str()),
Some(hm),
)
.instrument(tracing::info_span!("job_run", email = &email))
.await?;
}
if let Some(ref mut rsmq) = tx.rsmq {
rsmq.send_message(job_id.to_bytes_le().to_vec(), scheduled_for_o, tag);
}
Ok((uuid, tx))
}
pub fn canceled_job_to_result(job: &QueuedJob) -> serde_json::Value {
let reason = job
.canceled_reason
.as_deref()
.unwrap_or_else(|| "no reason given");
let canceler = job.canceled_by.as_deref().unwrap_or_else(|| "unknown");
serde_json::json!({"message": format!("Job canceled: {reason} by {canceler}"), "name": "Canceled", "reason": reason, "canceler": canceler})
}
async fn restarted_flows_resolution(
db: &Pool<Postgres>,
workspace_id: &str,
flow_value_if_any: Option<FlowValue>,
completed_flow_id: Uuid,
restart_step_id: &str,
branch_or_iteration_n: Option<usize>,
) -> Result<
(
Option<String>,
FlowValue,
i32,
Vec<FlowStatusModule>,
Option<i16>,
HashMap<String, serde_json::Value>,
FlowCleanupModule,
),
Error,
> {
let completed_job = sqlx::query_as::<_, CompletedJob>(
"SELECT *, null as labels FROM completed_job WHERE id = $1 and workspace_id = $2",
)
.bind(completed_flow_id)
.bind(workspace_id)
.fetch_one(db) // TODO: should we try to use the passed-in `tx` here?
.await
.map_err(|err| {
Error::InternalErr(format!(
"completed job not found for UUID {} in workspace {}: {}",
completed_flow_id, workspace_id, err
))
})?;
let raw_flow = completed_job
.parse_raw_flow()
.ok_or(Error::InternalErr(format!(
"Unable to parse raw definition for job {} in workspace {}",
completed_flow_id, workspace_id,
)))?;
let flow_status = completed_job
.parse_flow_status()
.ok_or(Error::InternalErr(format!(
"Unable to parse flow status for job {} in workspace {}",
completed_flow_id, workspace_id,
)))?;
let mut step_n = 0;
let mut dependent_module = false;
let mut truncated_modules: Vec<FlowStatusModule> = vec![];
for module in flow_status.modules {
if flow_value_if_any
.clone()
.map(|fv| {
fv.modules
.iter()
.find(|flow_value_module| flow_value_module.id == module.id())
.is_none()
})
.unwrap_or(false)
{
// skip module as it doesn't appear in the flow_value anymore
continue;
}
if module.id() == restart_step_id {
// if the module ID is the one we want to restart the flow at, or if it's past it in the flow,
// set the module as WaitingForPriorSteps as it needs to be re-run
if branch_or_iteration_n.is_none() || branch_or_iteration_n.unwrap() == 0 {
// The module as WaitingForPriorSteps as the entire module (i.e. all the branches) need to be re-run
truncated_modules.push(FlowStatusModule::WaitingForPriorSteps { id: module.id() });
} else {
// expect a module to be either a branchall (resp. loop), and resume the flow from this branch (resp. iteration)
let branch_or_iteration_n = branch_or_iteration_n.unwrap();
let module_definition = raw_flow
.modules
.iter()
.find(|flow_value_module| flow_value_module.id == restart_step_id)
.ok_or(Error::InternalErr(format!(
"Module {} not found in flow definition",
module.id()
)))?;
match module_definition.get_value() {
Ok(FlowModuleValue::BranchAll { branches, parallel, .. }) => {
if parallel {
return Err(Error::InternalErr(format!(
"Module {} is a parallel branchall. It can only be restarted at a given branch if it's sequential",
restart_step_id,
)));
}
let total_branch_number = module.flow_jobs().map(|v| v.len()).unwrap_or(0);
if total_branch_number <= branch_or_iteration_n {
return Err(Error::InternalErr(format!(
"Branch-all module {} has only {} branches. It can't be restarted on branch {}",
restart_step_id,
total_branch_number,
branch_or_iteration_n,
)));
}
let mut new_flow_jobs = module.flow_jobs().unwrap_or_default();
new_flow_jobs.truncate(branch_or_iteration_n);
truncated_modules.push(FlowStatusModule::InProgress {
id: module.id(),
job: new_flow_jobs[new_flow_jobs.len() - 1], // set to last finished job from completed flow
iterator: None,
flow_jobs: Some(new_flow_jobs),
branch_chosen: None,
branchall: Some(BranchAllStatus {
branch: branch_or_iteration_n - 1, // Doing minus one here as this variable reflects the latest finished job in the iteration
len: branches.len(),
}),
parallel: parallel,
while_loop: false,
});
}
Ok(FlowModuleValue::ForloopFlow { parallel, .. }) => {
if parallel {
return Err(Error::InternalErr(format!(
"Module {} is not parallel loop. It can only be restarted at a given iteration if it's sequential",
restart_step_id,
)));
}
let total_iterations = module.flow_jobs().map(|v| v.len()).unwrap_or(0);
if total_iterations <= branch_or_iteration_n {
return Err(Error::InternalErr(format!(
"For-loop module {} doesn't cannot be restarted on iteration number {} as it has only {} iterations",
restart_step_id,
branch_or_iteration_n,
total_iterations,
)));
}
let mut new_flow_jobs = module.flow_jobs().unwrap_or_default();
new_flow_jobs.truncate(branch_or_iteration_n);
truncated_modules.push(FlowStatusModule::InProgress {
id: module.id(),
job: new_flow_jobs[new_flow_jobs.len() - 1], // set to last finished job from completed flow
iterator: Some(Iterator {
index: branch_or_iteration_n - 1, // same deal as above, this refers to the last finished job
itered: vec![], // Setting itered to empty array here, such that input transforms will be re-computed by worker_flows
}),
flow_jobs: Some(new_flow_jobs),
branch_chosen: None,
branchall: None,
parallel: parallel,
while_loop: false,
});
}
_ => {
return Err(Error::InternalErr(format!(
"Module {} is not a branchall or forloop, unable to restart it at step {:?}",
restart_step_id,
branch_or_iteration_n
)));
}
}
}
dependent_module = true;
} else if dependent_module {
truncated_modules.push(FlowStatusModule::WaitingForPriorSteps { id: module.id() });
} else {
// else we simply "transfer" the module from the completed flow to the new one if it's a success
step_n = step_n + 1;
match module.clone() {
FlowStatusModule::Success {
id: _,
job: _,
flow_jobs: _,
branch_chosen: _,
approvers: _,
failed_retries: _,
} => Ok(truncated_modules.push(module)),
_ => Err(Error::InternalErr(format!(
"Flow cannot be restarted from a non successful module",
))),
}?;
}
}
if !dependent_module {
// step not found in flow.
return Err(Error::InternalErr(format!(
"Flow cannot be restarted from step {} as it could not be found.",
restart_step_id
)));
}
return Ok((
completed_job.script_path,
raw_flow,
step_n,
truncated_modules,
completed_job.priority,
flow_status.user_states,
flow_status.cleanup_module,
));
}
Reference in New Issue View Git Blame Copy Permalink